Workflow management is a technology that is considered strategically important by many businesses, and its market growth shows no signs of abating. It is, however, often viewed with skepticism by the research community, conjuring up visions of oppressed workers performing rigidly-defined tasks on an assembly line. Although the potential for abuse no doubt exists, workflow management can instead be used to help individuals manage their work and to provide a clear context for performing that work. A key challenge in the realization of this ideal is the reconciliation of workflow process models and software with the rich variety of activities and behaviors that comprise ``real'' work. Our experiences with the InConcert workflow management system are used as a basis for outlining several issues that will need to be addressed in meeting this challenge. This is intended as an invitation to CSCW researchers to influence this important technology in a constructive manner by drawing on research and experience.

This paper presents a study of Web content adaptation to improve server overload performance, as well as an implementation of a' Web content adaptation software prototype. When the request rate on a Web server increases beyond server capacity, the server becomes overloaded and unresponsive. The TCP listen queue of the server's socket overflows exhibiting a drop-tail behavior. As a result, clients experience service outages. Since clients typically issue multiple requests over the duration of a session with the server, and since requests are dropped indiscriminately, all clients connecting to the server at overload are likely to experience connection failures, even though there may be enough capacity on the server to deliver all responses properly for a subset of clients. In this paper, we propose to resolve the overload problem by adapting delivered content to load conditions to alleviate overload. The premise is that successful delivery of a less resource intensive content under overload is more desirable to clients than connection rejection or failures.

Today's Internet appliances feature user interface technologies almost unknown a few years ago: touch screens, styli, handwriting and voice recognition, speech synthesis, tiny screens, and more. This richness creates problems. First. different appliances use different languages: WML for cell phones; SpeechML, JSML, and VoxML for voice enabled devices such as phones; HTML and XUL for desktop computers, and so on. Thus, developers must maintain multiple source code families to deploy interfaces to one information system on multiple appliances. Second, user interfaces differ dramatically in complexity (e.g, PC versus cell phone interfaces). Thus, developers must also manage interface content. Third, developers risk writing appliance-specific interfaces for an appliance that might not be on the market tomorrow. A solution is to build interfaces with a single, universal language free of assumptions about appliances and interface technology. This paper introduces such a language, the User Interface Markup Language (UIML), an XML-compliant language. UIML insulates the interface designer from the peculiarities of different appliances through style sheets. A measure of the power of UIML is that it can replace hand-coding of Java AWT or Swing user interfaces.

Format: HTML Document(12K) . Audience: Non-technical, digital library researchers/funders. References: 13. Links: 2. Relevance: Low-medium. Abstract: Argues that social aspects of collaboration must be included in a Digital Library for the informal, organizational things that aren't always available in information sources. Mentions a TCL based system called CAFE that adds functionality of messages, bulletin boards, and talk.

Format: HTML Document(39K + pictures) .

Audience: Librarians, web masters. References: 27. Links: 2. Relevance: Low. Abstract: Discusses the problem of collection maintenance in the digital domain, and argues that while some traditional practices will carry over, new methods will have to be created, esp. for dynamic and informal resources. S

uggests that some maintenance can be done automatically by agents, and gives 2 examples: MOMSpider, which checks to make sure links are still current and Web:Lookout which notifies user when interesting changes are made to a watched

page.

Format: HTML Document(11K). Audience: Medical professionals,. References: 1. Links: 5. Relevance: None. Abstract: Describes the Visible Human Project (1 mm cross sections of two cadavers), how to obtain the images, how large they are, what IP agreements need to be signed.

The paper examines the research requirements of electronic commerce and digital libraries in six key areas. It provides case studies that describe three electronic commerce research projects (USC-ISI, CommerceNet, First Virtual) and six digital libraries projects sponsored by an NSF/ARPA/NASA initiative. The paper focuses on the following common areas of EC and DL research: acquiring and storing information; finding and filtering information; securing information and auditing access; universal access; cost management and financial instruments; and socio-economic impact.

The temporal elements of users’ information requirements are a continually confounding aspect of digital library design. No sooner have users’ needs been identified and supported than they change. This paper evaluates the changing information requirements of users through their ‘information journey’ in two different domains (health and academia). In-depth analysis of findings from interviews, focus groups and observations of 150 users have identified three stages to this journey: information initiation, facilitation (or gathering) and interpretation. The study shows that, although digital libraries are supporting aspects of users’ information facilitation, there are still requirements for them to better support users’ overall information work in context. Users are poorly supported in the initiation phase, as they recognize their information needs, especially with regard to resource awareness; in this context, interactive press-alerts are discussed. Some users (especially clinicians and patients) also required support in the interpretation of information, both satisfying themselves that the information is trustworthy and understanding what it means for a particular individual.

Format: HTML Document () .

Audience: Digital library researchers. References: 9. Links: 0. Relevance: Low. Abstract: Store pages as bitmaps, and retrieve a cite when user clicks on it, by doing OCR, then passing relevant line to library catalog, as 12 queries of 3 words each (randomly selected from the line) and returning the best scoring results. Somewhat robust to typos in cites, but not too slow.

Text documents often contain valuable structured data that is hidden in regular English sentences. This data is best exploited if available as a relational table that we could use for answering precise queries or for running data mining tasks. We explore a technique for extracting such tables from document collections that requires only a handful of training examples from users. these examples are used to generate extraction patterns, that in turn result in new tuples being extracted from the document collection. We build on this idea and present our Snowball system. Snowball introduces novel strategies for generating patterns and extracting tuples from plain-text documents. At each iteration of the extraction process, Snowball evaluates the quality of these patterns and tuples without human intervention, and keeps only the most reliable ones for the next iteration. In this paper we also develop a scalable evaluation methodology and metrics for our task, and present a thorough experimental evaluation of Snowball and comparable techniques over a collection for more than 300,000 newspaper documents.

The aim of this paper is to report the research results of an ongoing project that deals with the exploitation of a digital archive of drawings and illustrations of historic documents for research and education purposes. According to the results on a study of user requirements, we designed tools to provide researchers with novel ways for accessing the digital manuscripts, sharing, and transferring knowledge in a collaborative environment. Annotations are proposed for making explicit the results of scientific research on the relationships between images belonging to manuscripts produced in a time span of centuries. For this purpose, a taxonomy for linking annotation is proposed, together with a conceptual schema for representing annotations and for linking them to digital objects.

Modern cellular telephone systems extend the usability of portable personal computers enormously. A nomadic user can be given ubiquitous access to remote information stores and computing services. However, the behavior of wireless links creates severe inconveniences within the traditional data communication paradigm. We give an overview of the problems related to wireless mobility. We also present a new software architecture for mastering the problems and discuss a new paradigm for designing mobile distributed applications. The key idea in the architecture is to place a mediator, a distributed intelligent agent, between the mobile node and the wireline network.

Format: Not Yet Online.

Format: HTML Document (21K) .

Audience: non technical, users. References: 15. Links: 2. Relevance: Low. Abstract: Describes a particular user interface based on a book shelf metaphor. Tries to use an a priori classification (Dewey Decimal System) as an organization tool (in addition to results of electronic searches).

Format: HTML Document(33K + pictures) .

Audience: General Computer scientists, HCI . References: 22. Links: 1. Relevance: Low-Medium. Abstract: Uses metaphor of hierarchical Dewey Decimal system or faceted (implying a DAG) ACM literature categories to aid UI. Shows graphically where in the hierarchy hits were found for a search.

We introduce the idea of an ``event gazetteer'' that stores and presents locations in time. Each event is coded as a schema with attributes of event type, location, actor, and beginning and ending times. Sets of events can be collected as timelines and the events on these timelines can be linked by annotations. The system has been built with JSP and Oracle. Systematic metadata is essential for effective interaction with this system. For instance, the actors may be described by the roles in which they participate. In this paper, we focus on the construction of queries for this complex metadata. Ultimately, we envision a flexible, broad-based service that is a resource for users ranging from students to genealogists interested in events.

Events may be are best understood in the context of other events. Because of the temporal ordering, we can call a set of related events a timeline. Even such timelines are best understood in the context of other timelines. To facilitate the exploration of a collection of timelines and events, a visualization tool has been developed that structures the user's browsing. In this model, each event is accompanied by a text description and links to related resources. In particular, this system can provide a browsing interface of digitized historical newspapers.

Stories may be analyzed as sequences of causally-related events and reactions to those events by the characters. We employ a notation of plot elements, similar to one developed by Lehnert, and we extend that by forming higher level story threads. This notation requires that events and reactions be linked and that the chains of links be terminated back to the beginning of the story. Furthermore, we have built a browser for the plot elements, the story threads, and associated multimedia. We apply the browser to Corduroy, a children's short feature which was analyzed in detail. We provide additional illustrations with analysis of Kiss of Death, a Film Noir classic. Effectively, the browser provides a framework for interactive summaries of the narrative.

The WAP Web site from where the specs are available.

In this paper we analyze the behavior of e-commerce users based on actual logs from two large non-English e-brokers. We start by presenting a quantitative study of the behavior of e-brokers and discuss the influence of regional and cultural issues on them. We then discuss a model that quantifies the efficiency of the results provided by brokers in the electronic marketplace. This model is a function of factors such as server response time and regional factors. Our findings clearly indicate that e-commerce is strongly tied to local language, national customs and regulations, currency conversion and logistics, and Internet infrastructure. We found that the behavior of customers of online bookstores is strongly affected by brand and regional factors. Music CD shoppers show a different behavior that might stem from the fact that music is universal and not so language dependent.

Format: Compressed PostScript().

evaluating different link based ranking techniques

NCSTRL (Networked Computer Science Technical Reference Library) is a federation of digital libraries providing computer science materials. The architecture of the original NCSTRL was based largely on the Dienst software. It was implemented and maintained by the digital library group at Cornell University until September 2001. At that time, we had an immediate goal of preserving the existing NCSTRL collection and a long-term goal of providing a framework where participating organizations could continue to disseminate technical publications. Moreover, we wanted the new NCSTRL to be based on OAI (Open Archives Initiative) principles that provide a framework to facilitate the discovery of content in distributed archives. In this paper, we describe our experience in moving towards an OAI-based NCSTRL.

We describe two experimental desktop library clients that offer improved access to geospatial data via the Alexandria Digital Library (ADL): ArcADL, an extension to ESRI's ArcView GIS, and vtADL, an extension to the Virtual Terrain Project's Enviro terrain visualization package. ArcADL provides a simplified user interface to ADL's powerful underlying distributed geospatial search technology. Both clients use the ADL Access Framework to access library data that is available in multiple formats and retrievable by multiple methods. Issues common to both clients and future scenarios are also considered.

The Metis project is developing workflow technology designed for use in digital libraries by avoiding the assumptions made by traditional workflow systems. In particular, digital libraries have highly distributed sets of stakeholders who nevertheless must work together to perform shared activities. Hence, traditional assumptions that all members of a workflow belong to the same organization, work in the same fashion, or have access to similar computing platforms are invalid. The Metis approach makes use of event-based workflows to support the distributed nature of digital library workflow and employs techniques to make the resulting technology lightweight, flexible, and integrated with the Web. This paper describes the conceptual framework behind the Metis approach as well as a prototype which implements the framework. The prototype is evaluated based on its ability to model and execute a workflow drawn from a real-world digital library. After describing related work, the paper concludes with a discussion of future research opportunities in the area of digital library workflow and outlines how Metis is being deployed to a small set of digital libraries for additional evaluation.

An increasing number of systems, from pay-TV to electronic purses, rely on the tamper resistance of smartcards and other security processors. We describe a number of attacks on such systems some old, some new and some that are simply little known outside the chip testing community. We conclude that trusting tamper resistance is problematic; smartcards are broken routinely, and even a device that was described by a government signals agency as the most secure processor generally available' turns out to be vulnerable. Designers of secure systems should consider the consequences with care.

We present a hit inflation attack on pay-per- click Web advertising schemes. Our attack is virtually impossible for the program provider to detect conclusively, regardless of whether the provider is a third- party `ad network` or the target of the click itself. If practiced widely, this attack could accelerate a move away from pay- per-click program, and toward programs in which referrers are paid only if the referred user subsequently makes a purchase (pay-per-sale) or engages in other substantial activity at the target site (pay-per-lead). We also briefly discuss the lack of auditability inherent in these schemes.

We offer an overview of current Web search engine design. After introducing a generic search engine architecture, we examine each engine component in turn. We cover crawling, local Web page storage, indexing, and the use of link analysis for boosting search performance. The most common design and implementation techniques for each of these components are presented. We draw for this presentation from the literature, and from our own experimental search engine testbed. Emphasis is on introducing the fundamental concepts, and the results of several performance analyses we conducted to compare different designs.

Format: HTML Document(18K + pictures). Audience: computer scientists, digital library researchers. References: 1. Links: 3. Relevance: Medium-low. Abstract: Outlines 8 principles that are important to DLs, a combination of social/economic issues (avoid using words like ``copy'' and ``publish'') and technical ones (basically a sales pitch for the Kahn/Wilensky model of handles, maintenance, and access control.)

We describe an information seeking assistant for the world wide web. This agent, called WebWatcher, interactively helps users locate desired information by employing learned knowledge about which hyperlinks are likely to lead to the target information.

Format: Compressed PostScript().

Format: HTML Document (41K) . Audience: Scholars, publishers (esp. university press), librarians. References: 10. Links: 1. Relevance: Low-Medium. Abstract: Discusess the future of university presses, in pretty grim terms. Suggests that they lack the capital, staff, and quick reaction time to survive in an electronic world. Considers the Mellon report on scholarly comm

unication (which suggests universities get copyrights on books their faculty produce) unreasonable. Thinks that relying on commercial network providers (esp. cable, telecom) would be disastrous. Advocates a non-profit distribution ne

twork for scholarly publication.

Format: HTML Document (49K) . Audience: Librarians, policy makers. References: 10. Links: 1. Relevance: low. Abstract: A vision of the networked library. Sees the real value of librarians as creating attention structures which anticipate the way clients search.

The exchange of goods conducted face-to-face between two parties dates back to before the beginning of recorded history. Traditional means of payment have always had security problems, but now electronic payments retain the same drawbacks and add some risks. Unlike paper, digital documents can be copied perfectly and arbitrarily often, digital signatures can be produced by anybody who knows the secret cryptographic key, and a buyer's name can be associated with every payment, eliminating the anonymity of cash. Without new security measures, widespread electronic commerce is not viable. On the other hand, properly designed electronic payment systems can actually provide better security than traditional means of payments, in addition to flexibility. This article provides an overview of electronic payment systems, focusing on issues related to security.

The authors have recently re-implemented the Cedar language to make it portable across many different architectures. The strategy was, first, to use machine-dependent C code as an intermediate language, second, to create a language-independent layer known as the Portable Common Runtime, and third, to write a relatively large amount of Cedar-specific runtime code in a subset of Cedar itself. The paper presents a brief description of the Cedar language, the portability strategy for the compiler and runtime, the manner of making connections to other languages and the Unix operating system, and some performance measures of the Portable Cedar.

Unlike many efforts that focus on supporting scholarly research by developing large-scale, general resources for a wide range of audiences, we at the Cervantes Project have chosen to focus more narrowly on developing resources in support of ongoing research about the life and works of a single author, Miguel de Cervantes Saavedra (1547-1616). This has lead to a group of hypertextual archives, tightly integrated around the narrative and thematic structure of Don Quixote. This project is typical of many humanities research efforts and we discuss how our experiences inform the broader challenge of developing resources to support humanities research.

Format: HTML Document (43K) .

Audience: Semi-technical, business slant, funding proposal. References: 14. Links: 1. Relevance: Low. Abstract: Sees a DL as an enterprise wide collection of *executable* documents. SGML and Mathematica suggested as integration tools. Search for data representation which will allow automatic combination of separate documents to solve problems.

Rapid progress in hardware has led to the availability of portable personal computers ranging from laptops to hand-held computers (PDAs and Internet terminals). The presence of wireless connectivity gives these hand-held units the capability of accessing information anywhere, at any time. These mobile units can be considered to be part of a worldwide distributed information system. Distributed computing in mobile environments faces new challenges as more and more mobile hosts become an integral part of a distributed system. Problems in distributed computing in mobile environments are due to: (1) mobility, (2) wireless and (3) resource constraints at the mobile host. In this paper, we discuss the impact of these factors and research issues that need to be addressed in mobile distributed systems.

The chapters of the book are:

Introduction Modeling Retrieval Evaluation Query Languages (with Gonzalo Navarro) Query Operations Text and Multimedia Languages and Properties Text Operations (with Nivio Ziviani) Indexing and Searching (with Gonzalo Navarro) Parallel and Distributed IR (by Eric Brown) User Interfaces and Visualization (by Marti Hearst) Multimedia IR: Models and Languages (by Elisa Bertino, Barbara Catania and Elena Ferrari) Multimedia IR: Indexing and Searching (by Christos Faloutsos) Searching the Web Libraries and Bibliographic Systems (by Edie Rasmussen) Digital Libraries (by Edward Fox and Ohm Sornil) Appendix: Porter's Algorithm Glossary References (more than 800) Index

More information can be found in: http://www.sims.berkeley.edu/ hearst/irbook

Digital libraries of music have the potential to capture popular imagination in ways that more scholarly libraries cannot. we are working towards a comprehensive digital library of musical material, including popular music. We have developed new ways of collecting musical material, accessing it through searching and browsing, and presenting the results to the user. We work with different representations of music: facsimile images of scores, the internal representation of a music editing program, page images typeset by a music editor, MIDI files, audio files representing sung user input, and textual metadata such as title, composer and arranger, and lyrics. This paper describes a comprehensive suite of tools that we have built for this project. These tools gather musical material, convert between many of these representations, allow searching based on combined musical and textual criteria, and help present the results of searching and browsing. Although we do not yet have a single fully-blown digital music library, we have built several exploratory prototype collections of music, some of them very large (100,000 tunes), and critical components of the system have been evaluated.

People who create digital libraries need to gather together the raw material, add metadata as necessary, and design and build new collections. This paper sets out the requirements for these tasks and describes a new tool that supports them interactively, making it easy for users to create their own collections from electronic files of all types. The process involves selecting documents for inclusion, coming up with a suitable metadata set, assigning metadata to each document or group of documents, designing the form of the collection in terms of document formats, searchable indexes, and browsing facilities, building the necessary indexes and data structures, and putting the collection in place for others to use. All these tasks are supported within a modern point-and-click interaction paradigm. Although the tool is specific to the Greenstone digital library software, the underlying ideas should prove useful in more general contexts.

The World-Wide Web provides remote access to pages using its own naming scheme (URLs). transfer protocol (HTTP), and cache algorithms. Not only does using these special-purpose mechanisms have performance implications, but they make it impossible for standard Unix applications to access the Web. Gecko is a system that provides access to the Web via the NFS protocol. URLs are mapped to Unix file names, providing unmodified applications access to Web pages; pages are transferred from the Gecko server to the clients using NFS instead of HTTP. significantly improving performance; and NFS's cache consistency mechanism ensures that all clients have the same version of a page. Applications access pages as they would Unix files. A client-side proxy translates HTTP requests into file accesses, allowing existing Web applications to use Gecko. Experiments performed on our prototype show that Gecko is able to provide this additional functionality at a performance level that exceeds that of HTTP.

Traditional techniques for a distributed web server design rely on manipulation of central resources, such as routers or DNS services, to distribute requests designated for a single IP address to multiple web servers. The goal of the distributed cooperative Web server (DCWS) system development is to explore application-level techniques for distributing web content. We achieve this by dynamically manipulating the hyperlinks stored within the web documents themselves. The DCWS system effectively eliminates the bottleneck of centralized resources, while balancing the load among distributed web servers. DCWS servers may be located in different networks, or even different continents and still balance load effectively. DCWS system design is fully compatible with existing HTTP protocol semantics and existing web client software products.

The current exponential growth of the Internet precipitates a need for new tools to help people cope with the volume of information. To complement recent work on creating searchable indexex of the World-Wide Web and systems for filtering incoming e-mail and Usenet news articles, we describe a system which helps users keep abreast of new and interesting information. Every day it presents a selection of interesting web pages. The user evaluates each page, and given this feedback the system adapts and attempts to produce better pages the following day. We prsent some early results from an AI programming class to whom this was set as a project, and then describe our current implementation. Over the course of 24 days the output of our system was compared to both randomly-selected and human-selected pages. It consistently performed better than the random pages, and was better than the human-selected pages half of the time.

Online readers are in need of tools to help them cope with the mass of content that is available on the World Wide Web. In traditional media, readers are provided assistance in making selections. This includes both implicit assistance in the form of editorial oversight and explicit assistance in the form of recommendation services such as movie reviews and restaurant guides. The electronic medium offers new opportunities to create recommendation services, ones that adapt over time to track users' evolving interests. Fab is such a recommendation system for the Web, and has been operational in several versions since December 1994. By combining both collaborative and content-based filtering systems, Fab may eliminate many of the weaknesses found in each approach.

Format: Compressed PostScript

you give agent profile. It looks at the Web for things of interest and reports back. You give feedback

In a distributed, heterogeneous, proxy-based digital library, autonomous services and collections are accessed indirectly via proxies. To facilitate metadata compatibility and interoperability in such a digital library, we have designed a metadata architecture that includes four basic component classes: attribute model proxies, attribute model translators, metadata facilities for search proxies, and metadata repositories. Attribute model proxies elevate both attribute sets and the attributes they define to first-class objects. They also allow relationships among attributes to be captured. Attribute model translators map attributes and attribute values from one attribute model to another (where possible). Metadata facilities for search proxies provide structured descriptions both of the collections to which the search proxies provide access and of the search capabilities of the proxies. Finally, metadata repositories accumulate selected metadata from local instances of the other three component classes in order to facilitate global metadata queries and local metadata caching. In this paper, we outline further the roles of these component classes, discuss our design rationale, and analyze related work.

In a distributed, heterogeneous, proxy-based digital library, autonomous services and collections are accessed indirectly via proxies. To facilitate metadata compatibility and interoperability in such a digital library, we have designed a metadata architecture that includes four basic component classes: attribute model proxies, attribute model translators, metadata facilities for search proxies, and metadata repositories. Attribute model proxies elevate both attribute sets and the attributes they define to first-class objects. They also allow relationships among attributes to be captured. Attribute model translators map attributes and attribute values from one attribute model to another (where possible). Metadata facilities for search proxies provide structured descriptions both of the collections to which the search proxies provide access and of the search capabilities of the proxies. Finally, metadata repositories accumulate selected metadata from local instances of the other three component classes in order to facilitate global metadata queries and local metadata caching. In this paper, we outline further the roles of these component classes, discuss our design rationale, and analyze related work.

We describe an extensible constructor tool that helps information experts (e.g., librarians) create specialized query synthesizers for heterogeneous digital-library environments. A query synthesizer provides a graphical user interface in which a digital-library patron can specify a high-level, fielded, multi-source query. Furthermore, a query synthesizer interacts with a query translator and an attribute translator to transform high-level queries into sets of source-specific queries. We discuss how the constructor can facilitate discovery of available attributes (e.g., title), collation of schemas from different sources, selection of input widgets for a synthesizer (e.g., a text box or a drop-down list widget to support input of controlled vocabulary), and other design aspects. We also describe a prototype constructor we implemented, based on the Stanford InfoBus and metadata architecture.

Several ongoing Stanford University Digital Library projects address the issue of heterogeneity in networked information environments. A networked information environment has the following components: users, information repositories, information services, and payment mechanisms. This paper describes three of the heterogeneity-focused Stanford projects-InfoBus, REACH, and DLITE. The InfoBus project is at the protocol level, while the REACH and DLITE projects are both at the conceptual model level. The InfoBus project provides the infrastructure necessary for accessing heterogeneous services and utilizing heterogeneous payment mechanisms. The REACH project sets forth a uniform conceptual model for finding information in networked information repositories. The DLITE project presents a general task-based strategy for building user interfaces to heterogeneous networked information services.

We propose a user interaction model for browsing based on itera tive category-level operations. The motivation comes from two observations: 1) people naturally think in terms of categories, and 2) in browsing, the types of categories that are salient to users change as they browse. We define a set of category-level operations that lets users iteratively view and find results in terms of these changing category types. We also show that we can express some standard IR operations as iteratively applied sequences of a funda mental category-level operation (thus unifying them). Finally, we describe SenseMaker, a prototype interface for browsing heteroge neous sources.

Format: HTML Document () . Audience: Non-technical, on-line searchers. References: 16. Links: 1. Relevance: Low. Abstract: Addresses problem of finding correct keywords to search for by using WordNet. If a search doesn't turn up the hits needed, it modifies query by using synonyms, generalizing, or replacing with a set of more specific words. Searcher is asked to approve modified queries, which are then re-sent to content providers.

The rapid growth of the web has been noted and tracked extensively. Recent studies have however documented the dual phenomenon: web pages have small half lives, and thus the web exhibits rapid death as well. Consequently, page creators are faced with an increasingly burdensome task of keeping links up-to-date, and many are falling behind. In addition to just individual pages, collections of pages or even entire neighborhoods of the web exhibit significant decay, rendering them less effective as information resources. Such neighborhoods are identified only by frustrated searchers, seeking a way out of these stale neighborhoods, back to more up-to-date sections of the web; measuring the decay of a page purely on the basis of dead links on the page is too naive to reflect this frustration. In this paper we formalize a strong notion of a decay measure and present algorithms for computing it efficiently. We explore this measure by presenting a number of validations, and use it to identify interesting artifacts on today's web. We then describe a number of applications of such a measure to search engines, web page maintainers, ontologists, and individual users.

Streaming video on the World Wide Web is being widely deployed, and workplace training and distance education are key applications. The ability to annotate video on the Web can provide significant added value in these and other areas. Written and spoken annotations can provide `in context' personal notes and can enable asynchronous collaboration among groups of users. With annotations, users are no longer limited to viewing content passively on the Web, but are free to add and share commentary and links, thus transforming the Web into an interactive medium. We discuss design considerations in constructing a collaborative video annotation system, and we introduce our prototype, called MRAS. We present preliminary data on the use of Web- based annotations for personal note-taking and for sharing notes in a distance education scenario, Users showed a strong preference for MRAS over pen-and-paper for taking notes, despite taking longer to do so. They also indicated that they would make more abstract and questions with MRAS than in a `live' situation, and that sharing added substantial value.

In this paper, we describe five investment streams (data storage infrastructure, knowledge management, data production control, data transport and security, and personnel skill mix) that need to be balanced against short-term operating demands in order to maximize the probability of long-term viability of a digital library. Because of the rapid pace of information technology change, a digital library cannot be a static institution. Rather, it has to become a flexible organization adapted to continuous evolution of its infrastructure.

We present a new approach for modeling multi-modal data sets, focusing on the specific case of segmented images with associated text. Learning the joint distribution of image regions and words has many applications. We consider in detail predicting words associated with whole images (auto-annotation) and corresponding to particular image regions (region naming). Auto-annotation might help organize and access large collections of images. Region naming is a model of object recognition as a process of translating image regions to words, much as one might translate from one language to another. Learning the relationships between image regions and semantic correlates (words) is an interesting example of multi-modal data mining, particularly because it is typically hard to apply data mining techniques to collections of images. We develop a number of models for the joint distribution of image regions and words, including several which explicitly learn the correspondence between regions and words. We study multi-modal and correspondence extensions to Hofmann's hierarchical clustering/aspect model, a translation model adapted from statistical machine translation (Brown et al.), and a multi-modal extension to mixture of latent Dirichlet allocation (MoM-LDA). All models are assessed using a large collection of annotated images of real scenes. We study in depth the difficult problem of measuring performance. For the annotation task, we look at prediction performance on held out data. We present three alternative measures, oriented toward different types of task. Measuring the performance of correspondence methods is harder, because one must determine whether a word has been placed on the right region of an image. We can use annotation performance as a proxy measure, but accurate measurement requires hand labeled data, and thus must occur on a smaller scale. We show results using both an annotation proxy, and manually labeled data.

This pilot project investigates facilitating the development of the Semantic Web for e-learning through a practical example, using Materials Property Data Markup Language (MatML) to provide materials property data to a web-based application program. Property data for 100 materials is marked up with MatML and used as an input format for an application program. Students use the program to generate graphs showing selected properties for different materials. Selected graphs are submitted to the Materials Digital Library (MatDL) so that successive classes may be informed by earlier work to encourage new discoveries.

Format: Not yet online.

Today a mobile user wants to connect his portable computer: remotely to the central database at home, locally to the printer on the spot and globally to the world-wide-web. To achieve this, different connection lines are available: wireless networks for connecting out in the fields, ISDN or analogue telephone lines when residing in a hotel, Ethernet access at the customer's site. But this connectivity raises a lot of questions, about technical, security or accounting issues. This paper presents the architecture of an environment aiming to support mobile users and dealing with the given problems.

A comparison is made between information retrieval and information filtering. The authors determine that information filtering is a well defined process. By examining its foundations and comparing it to the foundations of the IR enterprise, the authors find there is very little difference between filtering and retrieval at an abstract level. They conclude that the two enterprises have the same goal; namely they are both concerned with getting information to people who need it. However, the authors emphasize that IR research has ignored some aspects of the general problem which both IR and information filtering address, and that these aspects are precisely those which especially relevant to the specific contexts of filtering.

Format: HTML Document (32K) . Audience: Semi-technical, general computer scientists. References: 8. Links: 1. Relevance: Medium (but not mainstream DL). Abstract: Discusses the interaction of compression and indexing. Suggests a Huffman encoding applied to words & non-words. Inverted bitmap for indexing, enhanced with Golomb encoding. Compressed 266 Mb Wall Street Journal a

rticle database by 50including creating the index. Queries were processed in less than .1 sec.

This paper proposes a family of protocols-iKP (i=1,2,3)-for secure electronic payments over the Internet. The protocols implement credit card-based transactions between the customer and the merchant while using the existing financial network for clearing and authorization. The protocols can be extended to apply to other payment models, such as debit cards and electronic checks. They are based on public-key cryptography and can be implemented in either software or hardware. Individual protocols differ in key management complexity and degree of security. It is intended that their deployment be gradual and incremental. The iKP protocols are presented herein with the intention to serve as a starting point for eventual standards on secure electronic payment.

This paper is focused on a central aspect in the design of our planned digital library for human movement, i.e. on the aspect of representation and recognition of human activity from video data. The method of representation is important since it has a major impact on the design of all the other building blocks of our system such as the user interface/query block or the activity recognition/storage block. In this paper we evaluate a representation method for human movement that is based on sequences of angular poses and angular velocites of the human skeletal joints, for storage and retrieval of human actions in video databases. The choice of a representation method plays an important role in the database structure, search methods, storage efficiency etc.. For this representation, we develop a novel approach for complex human activity recognition by employing multidimensional indexing combined with temporal or sequential correlation. this scheme is then evaluated with respect to its efficiency in storage and retrieval. For the indexing we use postures of humans in videos that are decomposed into a set of multidimensional tuples which represent the poses/velocities of human body parts such as arms, legs and torso. Three novel methods for human activity recognition are theoretically and experimentally compared. The methods require only a few sparsely sampled human postures. We also achieve speed invariant recognition of activities by eliminating the time factor and replacing it with sequence information. The indexing approach also provides robust recognition and an efficient storage/retrieval of all the activities in a small set of hash tables.

This paper proposes two enhancements to existing search services over the Web. One enhancement is the addition of limited dynamic search around results provided by regular Web search services, in order to correct part of the discrepancy between the actual Web and its static image as stored in search repositories. The second enhancement is an experimental two-phase paradigm that allows the user to distinguish between a domain query and a focused query within the dynamically identified domain. We present Fetuccino, an extension of the Mapuccino system that implements these two enhancements. Fetuccino provides an enhanced user-interface for visualization of search results, including advanced graph layout, display of structural information and support for standards (such as XML). While Fetuccino has been implemented on top of existing search services, its features could easily be integrated into any search engine for better performance. A light version of Fetuccino is available on the Internet at http://www.ibm.com/java/fetuccino.

The invention of the hyperlink and the HTTP transmission protocol caused an amazing new structure to appear on the Internet - the World Wide Web. With the Web, there came spiders, robots, and Web crawlers, which go from one link to the next checking Web health, ferreting out information and resources, and imposing organization on the huge collection of information (and dross) residing on the net. This paper reports on the use of one such crawler to synthesize document collections on various topics in science, mathematics, engineering and technology. Such collections could be part of a digital library.

Format: Not yet online.

TWO previous studies. one done at Stanford in 1997 based on data collected by the Google search engine, and one done at Digital in 1996 based on AltaVista data, revealed that almost a third of the Web consists of duplicate pages. Both studies identified mirroring, that is, the systematic replication of content over a pair of hosts, as the principal cause of duplication, but did not further investigate this phenomenon. The main aim of this paper is to present a clearer picture of mirroring on the Web. As input we used a set of 179 million URLs found during a Web crawl done in the summer of 1998. We looked at all hosts with more than 100 URLs in our input (about 238,000), and discovered that about 10the prevalence of mirroring based on a mirroring classification scheme that we define. There are numerous reasons for mirroring: technical (e.g., to improve access time), commercial (e.g., different intermediaries offering the same products), cultural (e.g., same content in two languages), social (e.g.. sharing of research data). and so forth. Although we have not done a exhaustive study of the causes of replication, we discuss and provide examples for several representative cases. Our technique for detecting mirrored hosts from large sets of collected URLs depends mostly on the syntactic analysis of URL strings, and requires retrieval and content analysis only for a small number of pages. We are able to detect both partial and total mirroring, and handle cases where the content is not byte-wise identical. Furthermore, our technique is computationally very efficient and does not assume that the initial set of URLs gathered from each host is comprehensive. Hence, this approach has practical uses beyond our study, and can be applied in other settings. For instance, for Web crawlers and caching proxies, detecting mirrors can be valuable to avoid redundant fetching. and knowledge of mirroring can be used to compensate for broken links.

The authors discuss an object oriented approach to network management. Their goal is to briefly explain a real example of an integrated network management (INM) system. One of the major requirements when looking at information transfer between the managed network and the management system is to mask the heterogeneity of the underlying resources. As an example of the unification of heterogeneity networks, a software called the Integrator has been designed and implemented. The Integrator is a mechanism that provides an object oriented interface to the user (human or network management application programs) to offer a homogeneous view of a world (set of heterogeneous domains) through a model (depicting a formal information view). The Integrator uses two agents to communicate with underlying network elements: an SNMP agent accessing TCP/IP parameters for an Ethernet network through a SNMP agent, and an X.25 interface program doing the same for X.25 parameters through proprietary management software. The concepts of the Integrator has been applied in the EC project PEMMON

Software tools, including Web browsers, e-books, electronic document formats, search engines, and digital libraries are changing the way that people read, making it easier for them to find and view documents. However, while these tools provide significant help with short-term reading projects involving small numbers of documents, they fall short of supporting readers engaged in longer-term reading projects, in which a topic is to be understood in-depth by reading many documents. Such readers need to find and manage many documents and citations, remember what they have read, and prioritize what to read next. In this paper, we describe three integrated software tools that facilitate in-depth reading. A first tool extracts citation information from documents. A second finds on-line documents from their citations. The last is a document corpus browser that uses a zoomable user interface to show a corpus at multiple granularities while supporting reading tasks that take days, weeks, or longer. We describe these tools and the design principles that motivated them.

We present two techniques that make document collection visualizations more informative. Icon abacus uses the horizontal position of icon groups to communicate document attributes. Ghost icons show linked documents by adding temporary icons and by highlighting or dimming existing ones.

Format: HTML Document().

Format: HTML Document (36K) . Audience: slightly technical, generalist comfortable with technology, funders. References: 13. Links: 1. Relevance: Medium-High. Abstract: Describes the UMichigan Digital Libraries proposal, including some detail about their agent architecture. User agents, Collection-interface agents, and mediators all play a role. Network resources are allocated on a market-based mechanism, and proposal mentions need to protect intellectual property & handle payment issues.

Format: Compressed PostScript().

Format: HTML Document().

A new federal initiative called Information Technology for the Twenty-First Century (IT2) recognizes the need to bridge research across domains in or&r to bring computing benefits to society at large. One implication for digital library (DL) research is that we should start looking at projects that span the spectrum from basic computer science to the implementation of working systems and consider links among findings on information system use from a variety of arenas in life. In this paper, I integrate findings from my research on people's encounters with DLs in two different arenas: academia and low-income neighborhoods. The point is to see how concepts and conclusions related to use do, in fact, cross these arenas. The paper also aims to help bring results from studies of local community information practices into the realm of DLs, since community networking represents one particular genre and audience that has not yet received a great deal of attention from those engaged in DL research. Beginning with a discussion of DL use as an assemblage of infrastructure, norms, knowledge, and practice, the paper explores a number of insights gleaned from user studies associated with two separate research projects: 1) the recently completed University of Illinois Digital Libraries Initiative (DLI) project; and 2) the Community Networking Initiative (CNI) currently in progress under the auspices of the University of Illinois, the Urban League of Champaign County and Prairienet, the community network serving East Central Illinois. Insights about DL use discussed in this paper include: the way in which trivial barriers are magnified until they effectively cut off use on a large scale; the difficulties faced by outsiders whose information worlds are impoverished, the primacy of comfort and relevant content in encouraging use; and the importance of informal social networks for providing help related to system use.

In the race to transform the World Wide Web from a medium for information presentation to a medium for information exchange, the development of practices for ensuring the security, auditability, and non- repudiation of transactions that are well established in the paper-based world has not kept pace in the digital world. Existing Internet technology provides no easy way to create a valid `digital receipt' that meets the requirements of both complex distributed networks and the business community. In addition, an improved articulation of digital signatures is needed. Extensible Forms Description Language (XFDL), developed by UWI.Com and Tim Bray, is an application of XML that allows organizations to move their paper-based forms systems to the Internet while maintaining the necessary attributes of paper-based transaction records. XFDL was designed for implementation in business-to-business electronic commerce and intra-organizational information transactions.

Advances in both technology and publishing practices continue to increase the quantity of scientific literature that is available electronically. In this paper, we introduce the Information Synthesis process, a new approach that enables scientists to visualize, explore, and resolve contradictory findings that are inevitable when multiple empirical studies explore the same natural phenomena. Central to the Information Synthesis approach is a cyber-infrastructure that provides a scientist with both secondary information from an article and structured information resources. To demonstrate this approach, we have developed the Multi-User, Information Extraction for Information Synthesis (METIS) System. METIS is an interactive system that automates critical tasks within the Information Synthesis process. We provide two case-studies that demonstrate the utility of the Information Synthesis approach.

The authors report their experiences building the query optimizer for TI's Open OODB system. It is probably the first working object query optimizer to be based on a complete extensible optimization framework including logical algebra, execution algorithms, property enforcers, logical transformation rules, implementation rules, and selectivity and cost estimation. Their algebra incorporates a new materialize operator with its corresponding logical transformation and implementation rules that enable the optimization of path expressions. The Open OODB query optimizer was constructed using the Volcano Optimizer Generator, demonstrating that this second-generation optimizer generator enables rapid development of efficient and effective query optimizers for non-standard data models and systems.

There are two main kinds of approach to considering usability of any system: empirical and analytical. Empirical techniques involve testing systems with users, whereas analytical techniques involve usability personnel assessing systems using established theories and methods. We report here on a set of studies in which four different techniques were applied to various digital libraries, focusing on the strengths, limitations and scope of each approach. Two of the techniques, Heuristic Evaluation and Cognitive Walkthrough, were applied in text-book fashion, because there was no obvious way to contextualize them to the Digital Libraries (DL) domain. For the third, Claims Analysis, it was possible to develop a set of re-usable scenarios and personas that relate the approach specifically to DL development. The fourth technique, CASSM, relates explicitly to the DL domain by combining empirical data with an analytical approach. We have found that Heuristic Evaluation and Cognitive Walkthrough only address superficial aspects of interface design (but are good for that), whereas Claims Analysis and CASSM can help identify deeper conceptual difficulties (but demand greater skill of the analyst). However, none fit seamlessly within the fragmented function-oriented design practices that typify much digital library development, highlighting an important area for further work to support improved usability.

The aim of the work reported here was to better understand the usability issues raised when digital libraries are used in a natural setting. The method used was a protocol analysis of users working on a task of their own choosing to retrieve documents from publicly available digital libraries. Various classes of usability difficulties were found. Here, we focus on use in context - that is, usability concerns that arise from the fact that libraries are accessed in particular ways, under technically and organisationally imposed constraints, and that use of any particular resource is discretionary. The concepts from an Interaction Framework, which provides support for reasoning about patterns of interaction between users and systems, are applied to understand interaction issues.

Format: Not Yet Online.

We introduce a system as part of the CiteSeer digital library project for automatic tracking of scientific literature that is relevant to a user's research interests. Unlike previous systems that use simple keyword matching, CiteSeer is able to track and recommend topically relevant papers even when keyword based query profiles fail. This is made possible through the use of a heterogenous profile to represent user interests. These profiles include several representations, including content based relatedness measures. The CiteSeer tracking system is well integrated into the search and browsing facilities'of CiteSeer, and provides the user with great flexibility in tuning a profile to better match his or her interests. The software for this system is available, and a sample database is online as a public service.

Format: HTML Document(21K + pictures)

Audience: Calculus Instructors. References: 13. Links: 16. Abstract: Discusses an architecture for a system that aids in the teaching of calculus.

As the development of pen computing' continues, more and more of today's computers are likely gradually to move away from people's desktops and into their pockets. The development of personal digital assistants (PDAs) has initiated this move. As these devices move into people's pockets, they need the ability to access information on the move. This article describes a generic view of a client server mobile computing architecture. It also sheds some light on the basic network topologies that have been considered previously for such systems. The scenario used is a hospital ward. Each doctor is equipped with a PDA and each ward or a group of wards with a server providing patient records. As a doctor visits a patient in a ward, the patient's record is accessed from the server onto the PDA. The doctor updates the record and sends the update back to the server.

This paper describes the architecture and components of the infrastructure in construction for the National Digital Library in Portugal. The requirements emerged from the definition of the services to support, with a special focus on scalability, and from the decision to give a special attention to community building standards, open solutions, and reusable and cost effective components. The generic bibliographic metadata format in this project is UNIMARC, and the structural metadata is METS. The URN identifiers are processed and resolved as simple but very effective PURL identifiers, and the storage is provided by the new emerging LUSTRE file system, for immediate access, and by a locally developed GRID architecture, ARCO, for long term preservation. All these components run on Linux servers, as also the middleware for access based in the FEDORA framework.

Describes collab-related aspects of Andrew.

A goal of the Alexandria Digital Earth Prototype (ADEPT) project is to make primary resources in geography useful for undergraduate instruction in ways that will promote inquiry learning. The ADEPT education and evaluation team interviewed professors about their use of geography information as they prepare for class lectures, as compared to their research activities. We found that professors desired the ability to search by concept (erosion, continental drift, etc.) as well as geographic location, and that personal research collections were an important source of instructional materials. Resources in geo-spatial digital libraries are typically described by location, but are rarely described by concept or educational application. This paper presents implications for the design of an educational digital library from our observations of the lecture preparation process. Findings include functionality requirements for digital libraries and implications for the notion of digital libraries as a shared information environment. The functional requirements include definitions and enhancements of searching capabilities, the ability to contribute and to share personal collections of resources, and the capability to manipulate data and images.

This paper argues for the design of collaborative visual interfaces to digital libraries that support social navigation. As an illustrative example we present work in progress on the design of a three-dimensional document space for a scholarly community - namely faculty, staff, and students at the School of Library and Information Science, Indiana University. We conclude with a set of research challenges.

It is increasingly difficult to make effective use of Internet information, given the rapid growth in data volume, user base, and data diversity. We introduce Harvest, a system that provides a scalable, customizable architecture for gathering, indexing, caching, replicating, and accessing Internet information.

Interactive Web services are increasingly replacing traditional static Web pages. Producing Web services seems to require a tremendous amount of laborious low-level coding due to the primitive nature of CGI programming. We present ideas for an improved runtime system for interactive Web services built on top of CGI running on virtually every combination of browser and HTTP/CGI server. The runtime system has been implemented and used extensively in <bigwig>. a tool for producing interactive Web services.

In this paper we study how to make web servers (e.g., Apache) more crawler friendly. Current web servers offer the same interface to crawlers and regular web surfers, even though crawlers and surfers have very different performance requirements. We evaluate simple and easy-to-incorporate modifications to web servers so that there are significant bandwidth savings. Specifically, we propose that web servers export meta-data archives decribing their content.

Web searching and browsing can be improved if browsers and search engines know which pages users frequently visit. 'Web tracking' is the process of gathering that information. The goal for Web tracking is to obtain a database describing Web page download times and users' page traversal patterns. The database can then be used for data mining or for suggesting popular or relevant pages to other users. We implemented three Web tracking systems, and compared their performance. In the first system, rather than connecting directly to Web sites, a client issues URL requests to a proxy. The proxy connects to the remote server and returns the data to the client, keeping a log of all transactions. The second system uses sniffers to log all HTTP traffic on a subnet. The third system periodically collects browser log files and sends them to a central repository for processing. Each of the systems differs in its advantages and pitfalls. We present a comparison of these techniques.

Format: Not yet online.

Format: Not yet online.

This article presents a taxonomy of global information-sharing systems and discusses where multidatabase systems fit in the spectrum of solutions. The authors use this taxonomy as a basis for defining multidatabase systems, then discuss the issues associated with them. In particular, the paper focuses on two major design approaches- global schema systems and multidatabase language systems.

In this paper, we present Google, a prototype of a large-scale search engine which makes heavy use of the structure present in hypertext. Google is designed to crawl and index the Web efficiently and produce much more satisfying search results than existing systems. The prototype with a full text and hyperlink database of at least 24 million pages is available at http://google.stanford.edu/ To engineer a search engine is a challenging task. Search engines index tens to hundreds of millions of web pages involving a comparable number of distinct terms. They answer tens of millions of queries every day. Despite the importance of large-scale search engines on the web, very little academic research has been done on them. Furthermore, due to rapid advance in technology and web proliferation, creating a web search engine today is very different from three years ago. This paper provides an in-depth description of our large-scale web search engine - the first such detailed public description we know of to date. Apart from the problems of scaling traditional search techniques to data of this magnitude, there are new technical challenges involved with using the additional information present in hypertext to produce better search results. This paper addresses this question of how to build a practical large-scale system which can exploit the additional information present in hypertext. Also we look at the problem of how to effectively deal with uncontrolled hypertext collections where anyone can publish anything they want.

In a digital library system, documents are available in digital form and therefore are more easily copied and their copyrights are more easily violated. This is a very serious problem, as it discourages owners of valuable information from sharing it with authorized users. There are two main philosophies for addressing this problem: prevention and detection. The former actually makes unauthorized use of documents difficult or impossible while the latter makes it easier to discover such activity. We propose a system for registering documents and then detecting copies, either complete copies or partial copies. We describe algorithms for such detection, and metrics required for evaluating detection mechanisms (covering accuracy, efficiency, and security). We also describe a working prototype, called COPS, describe implementation issues, and present experimental results that suggest the proper settings for copy detection parameters.

Seed a search with examples of a pattern, such as citations to books. Let the engine run over Web pages and learn. Get back more books.

Shows architecture of Google.

Describes a new architecture for data mining. It makes use of some of the dynamic itemset counting technology

Significant efforts are being made to digitize rare and valuable library materials, with the goal of providing patrons and historians digital facsimiles that capture the look and feel of the original materials. This is often done by digitally photographing the materials and making high resolution 2D images available. The underlying assumption is that the objects are flat. However, older materials may not be flat in practice, being warped and crinkled due to decay, neglect, accident and the passing of time. In such cases, 2D imaging is insufficient to capture the look and feel of the original. For these materials, 3D acquisition is necessary to create a realistic facsimile. This paper outlines a technique for capturing an accurate 3D representation of library materials which can be integrated directly into current digitization setups. This will allow digitization efforts to provide patrons with more realistic digital facsimile of library materials.

This paper presents research focused on developing new techniques and algorithms for the digital acquisition, restoration, and study of damaged manuscripts. We present results from an acquisition effort in partnership with the British Library, funded through the NSF DLI-2 program, designed to capture 3-D models of old and damaged manuscripts. We show how these 3-D facsimiles can be analyzed and manipulated in ways that are tedious or even impossible if confined to the physical manuscript. In particular, we present results from a restoration framework we have developed for flattening the 3-D representation of badly warped manuscripts. We expect these research directions to give scholars more sophisticated methods to preserve, restore, and better understand the physical objects they study.

We present a technique to correct image distortion that can occur when library materials are imaged by cameras. Our approach provides a general framework to undo a variety of common distortions, including binder curl, fold distortion, and combinations of the two. Our algorithm is described and demonstrated on several examples.

Format: HTML Document().

Format: HTML Document (35K + pictures) .

Audience: Computer scientists, mathematicians, librarians. References: 15. Links: 13. Relevance: Low-Medium. Abstract: Describes the NHSE software repository, with files kept at authors' sites, but a central index in a common form (prepared manually now, but hopefully automatically later). Includes a special process for submission a

nd revision via web forms, with digital signatures (PGP) required for authentication. Accepted files are fingerprinted using MD5 so that modifications can be detected. A scheme of LIFNs (Location Independent FileNames) is essentially

a precursor to URN's.

This paper explores two ways to help students locate most relevant resources in educational digital libraries. One is a more comprehensive access to educational resources through several ways of information access including browsing and information visualization. Another is personalized information access through social navigation support. The paper presents the details of the Knowledge Sea III system for comprehensive personalized access to educational resources and presented results of a classroom study. The study delivered a convincing argument for the importance of providing several ways of information showing that only about 10% of all resource accesses were made through the traditional search interface. We have also collected some good evidence in favor of the social navigation support.

This paper introduces a new framework for building digital library collections and contrasts it with existing systems. It describes a significant new step in the development of a widely-used open-source digital library system, Greenstone, which has evolved over many years. It is supported by a fresh implementation, which forced us to rethink the entire design rather than making incremental improvements. The redesign capitalizes on the best ideas from the existing system, which have been refined and developed to open new avenues through which digital librarians can tailor their collections. We demonstrate its flexibility by showing how digital library collections can be extended and altered to satisfy new requirements.

Users of modern digital libraries (DLs) can keep themselves up-to-date by searching and browsing their favorite collections, or more conveniently by resorting to an alerting service. The alerting service notifies its clients about new or changed documents. So far, no sophisticated service has been proposed that covers heterogeneous and distributed collections and is integrated with the digital library software.

This paper analyses the conceptual requirements of this much-sought after service for digital libraries. We demonstrate that the diffing concepts of digital libraries and its underlying technical design has extensive influence (a) the expectations, needs and interests of users regarding an alerting service, and (b) on the technical possibilities of the implementation of the service.

Our findings will show that the range of issues surrounding alerting services for digital libraries, their design and use is greater than one may anticipate. We also show that, conversely, the requirements for an alerting service have considerable impact on the concepts of DL design. Our findings should be of interest for librarians as well as system designers. We highlight and discuss the far-reaching implications for the design of, and interaction with, libraries. This paper discusses on the lessons learned from building such a distributed alerting service. We present our prototype implementation as a proof-of-concept for an alerting service for open DL software.

Format: Not Yet On-line.

Audience: Technical. HyTime developers.. References: 17. Links: . Relevance: Low. Abstract: HyTime is an ISO standard for hypermedia time based documents. This paper discusses an implementation of a database and search engine operating in that language. Examples of queries, optimizations, etc.

The Electronic Meeting Assistance (EMA) is a virtually co-located mixed system. That means that all participants of the meeting are present at the same time but not necessary at the same location (for example some people meet in a room and an external, remote expert is included via a local area network). During the meeting the personal Notepads of the participants are linked together using a radio LAN. In addition an interactive white-board e.g. the Xerox LiveBoard is used for visualisation and manipulation of common data. To assist cooperative work, the EMA system supports the exchange of information during meetings. Various information can be exchanged between meeting members, for example contact information, prepared notes and diagrams; electronic presentations could be given or a paper could be edited by the group.

Format: Compressed PostScript().

A specific task of distributed and parallel information systems is workflow management. In particular, workflow management systems execute business processes that run on top of distributed and parallel information systems. Parallelism is due to performance requirements and involves data and applications that are spread across a heterogeneous, distributed computing environment. Heterogeneity and distribution of the underlying computing infrastructure should be made transparent in order to alleviate programming and use. We introduce an implementation architecture for workflow management systems that meets these requirements. Scalability (through transparent parallelism) and transparency with respect to distribution and heterogeneity are the major characteristics of this architecture. A generic client/server class library in an object-oriented environment demonstrates the feasibility of the approach.

Format: Compressed PostScript().

We demonstrate a new browsing technique for devices with small displays such as PDAs or cellular phones. We concentrate on end-game browsing, where the user is close to or on the target page. We make browsing more efficient and easier by Accordion Summarization. In this technique the Web page is first represented as a short summary. The user can then drill down to discover relevant parts of the page. If desired, keywords can be highlighted and exposed automatically. We discuss our techniques, architecture, interface facilities, and the result of user evaluations. We measured a 57 improvement in browsing speed and 75 input effort.

The Stanford Power Browser project addresses the problems of interacting with the World-Wide Web through wirelessly connected Personal Digital Assistants (PDAs). These problems include bandwidth limitations, screen real-estate shortage, battery capacity, and the time costs of pen-based search keyword input. As a way to address bandwidth and battery life limitations, we provide local site search facilities for all sites. We incrementally index Web sites in real time as the PDA user visits them. These indexes have narrow scope at first, and improve as the user dwells on the site, or as more users visit the site over time. We address the keyword input problem by providing site specific keyword completion, and indications of keyword selectivity within sites. The system is implemented on the Palm Pilot platform, using a Metricom radio link. We describe the user level experience, and then present the analyses that informed our technical decisions.

We introduce five methods for summarizing parts of Web pages on handheld devices, such as personal digital assistants (PDAs), or cellular phones. Each Web page is broken into text units that can each be hidden, partially displayed, made fully visible, or summarized. The methods accomplish summarization by different means. One method extracts significant keywords from the text units, another attempts to find each text unit's most significant sentence to act as a summary for the unit. We use information retrieval techniques, which we adapt to the World-Wide Web context. We tested the relative performance of our five methods by asking human subjects to accomplish single-page information search tasks using each method. We found that the combination of keywords and single-sentence summaries works best for a variety of search tasks.

We introduce five methods for summarizing parts of Web pages on handheld devices, such as personal digital assistants (PDAs), or cellular phones. Each Web page is broken into text units that can each be hidden, partially displayed, made fully visible, or summarized. The methods accomplish summarization by different means. One method extracts significant keywords from the text units, another attempts to find each text unit's most significant sentence to act as a summary for the unit. We use information retrieval techniques, which we adapt to the World-Wide Web context. We tested the relative performance of our five methods by asking human subjects to accomplish single-page information search tasks using each method. We found that the combination of keywords and single-sentence summaries works best for a variety of search tasks.

We have designed and implemented new Web browsing facilities to support effective navigation on Personal Digital Assistants (PDAs) with limited capabilities: low bandwidth, small display, and slow CPU. The implementation supports wireless browsing from 3Com's Palm Pilot. An HTTP proxy fetches web pages on the client's behalf and dynamically generates summary views to be transmitted to the client. These summaries represent both the link structure and contents of a set of web pages, using information about link importance. We discuss the architecture, user interface facilities, and the results of comparative performance evaluations. We measured a 45 and a 42% reduction in required pen movements.

Many information sources on the web are relevant primarily to specific geographical communities. For instance, web sites containing information on restaurants, theatres and apartment rentals are relevant primarily to web users in geographical proximity to these locations. We make the case for identifying and exploiting the geographical location information of web sites so that web applications can rank information in a geographically sensitive fashion. For instance, when a user in Palo Alto issues a query for Italian Restaurants, a web search engine can rank results based on how close such restaurants are to the user's physical location rather than based on traditional IR measures. In this paper, we first consider how to compute the geographical location of web pages. Subsequently, we consider how to exploit such information in one specific proof-of-concept application we implemented in JAVA.

We are interested in questions of improving user control in best- match text-retrieval systems, specifically questions as to whether simple visualizations that nonetheless go beyond the minimal ones generally available can significantly help users. Recently, we have been investigating ways to help users decide-given a set of documents retrieved by a query-which documents and passages are worth closer examination. We built a document viewer incorporating a visualization centered around a novel content-displaying scrollbar and color term highlighting, and studied whether the visualization is helpful to non-expert searchers. Participants' reaction to the visualization was very positive, while the objective results were inconclusive.

Almost all work on music information retrieval to date has concentrated on music in the audio and event (normally MIDI) domains. However, music in the form of notation, expecially Conventional Music Notation (CMN), is of much interest to musically-trained persons, both amateurs and professionals, and searching CMN has great value for digital music libraries. One obvious reason little has been done on music retrieval in CMN form is the overwhelming complexity of CMN, which requires a very substantial investment in programming before one can even begin studying music IR. This paper reports on work adding music-retrieval capabilities to Nightingale, an existing professional-level music-notation editor.

The Variations2 digital music library currently supports music in audio and score-image formats. In a future version, we plan to add music in a symbolic form. This paper describes our work defining a music representation suitable for the needs of our users.

Link Analysis has shown great potential in improving the performance of web search. PageRank and HITS are two of the most popular algorithms. Most of the existing link analysis algorithms treat a web page as a single node in the web graph. However, in most cases, a web page contains multiple semantics and hence the web page might not be considered as the atomic node. In this paper, the web page is partitioned into blocks using the vision-based page segmentation algorithm. By extracting the page-to-block, block-to-page relationships from link structure and page layout analysis, we can construct a semantic graph over the WWW such that each node exactly represents a single semantic topic. This graph can better describe the semantic structure of the web. Based on block-level link analysis, we proposed two new algorithms, Block Level PageRank and Block Level HITS, whose performances we study extensively using web data.

The Web contains a huge volume of unstructured data, which is difficult to manage. In digital libraries, on the other hand, information is explicitly organized, described, and managed. Community-oriented services are built to attend specific information needs and tasks. In this paper, we describe an environment, Web-DL, that allows the construction of digital libraries from the Web. The Web-DL environment will allow us to collect data from the Web, standardize it and publish it through a digital library system. It provides support to services and organizational structure normally available in digital libraries, but benefiting from the breadth of the Web contents. We experimented with applying the Web-DL environment to the Networked Digital Library of Theses and Dissertations (NDLTD), thus demonstrating that the rapid construction of DLs from the Web is possible. Also, Web-DL provides an alternative as a large-scale solution for interoperability between independent digital libraries.

In this paper, we introduce a multifaceted approach for question answering on lecture videos. Text extracted from PowerPoint slides associated with the lecture videos is used as a source of domain knowledge to boost the answer extraction performance on these domain specific videos. The three steps of this approach are described and the evaluation plan is discussed.

This is a comprehensive book on Database security. Chapter 1,2, and 3 describe information security, security models and security mechanisms and software from a general point of view. Chapter 4 gives a detail survey of Database security design. Chapter 5 explores the problem of security on statistical databases. Chapeter 6 describes different approaches in instrusion detection. Chapter 7 explores security models for next-generation databases (active db, oodb).

Mobile devices represent new opportunities for accessing digital libraries (DLs) but also pose a number of challenges given the diversity of their hardware and software features. We describe a framework aimed at facilitating the generation of interfaces for access to DL resources from a wide range of mobile devices.

Expandability is one of the main requirements of future digital libraries. This paper introduces a digital library service system, OpenDLib, that has been designed to be highly expandable both in terms of content, services and usage. The paper illustrate the mechanisms that enable expandability and discusses their impact on the development of the system architecture.

Format: HTML Document (9K) .

Audience: Semi-technical, general computer science. References: 5. Links: 1. Relevance: Low. Abstract: Discusses need for inexact matching, eg. OCR recognition errors. Proposes using N-grams, overlapping sequences of N adjacent letters as search target. Also research in matching in image of scanned documents (not doing OCR). Some results on mail sorting & census data.

In the office environment, the retrieval of documents is performed using the concepts contained in the documents, information about the procedural context where the documents are used, and information about the regulations and laws that discipline the life of documents within a given application domain. To fulfill the requirements of such a sophisticated retrieval, we propose a document retrieval model and system based on the representation of knowledge describing the semantic contents of documents, the way in which the documents are managed by procedures and by people in the office, and the application domain where the office operates. The article describes the knowledge representation issues needed for the document retrieval system and presents a document retrieval model that captures these issues. The effectiveness of the approach is illustrated by describing a system, named Kabiria, built on top of such model. The article describes the querying and browsing environments, and the architecture of the system.

The growing acceptance of XML as a standard for semi-structured documents on the Web opens up challenging opportunities for Web query languages. In this paper we introduce XML-GL, a graphical query language for XML documents. The use of a visual formalism for representing both the content of XML documents (and of their DTDs) and the syntax and semantics of queries enables an intuitive expression of queries, even when they are rather complex. XML-GL is inspired by G-log, a general purpose, logic-based language for querying structured and semi-structured data. The paper presents the basic capabilities of XML-GL through a sequence of examples of increasing complexity.

Textbook

A musical piece typically has a repetitive structure. Analysis of this structure will be useful for music segmentation, indexing and thumbnailing. This paper presents an algorithm that can automatically analyze the repetitive structure of musical signals. First, the algorithm detects the repetition of each segment of fixed length in a piece using dynamic programming. Second, the algorithm summarizes this repetition information and infers the structure based on heuristic rules. The performance of the approach is demonstrated visually using figures for qualitative evaluation, and by two structural similarity measures for quantitative evaluation. Based on the structural analysis result, this paper also proposes a method for music thumbnailing. The preliminary results obtained using a corpus of BeatlesÆ songs show that automatic structural analysis and thumbnailing of music are possible.

From a user's perspective, hypertext links on the Web form a directed graph between distinct information sources. We investigate the effects of discovering `backlinks' from Web resources, namely links pointing to the resource. We describe fools for backlink navigation on both the client and server side, using an applet for the client and a module for the Apache Web server, We also discuss possible extensions to the HTTP protocol to facilitate the collection and navigation of backlink information in the World Wide Web.

The rapid growth of the World-Wide Web poses unprecedented scaling challenges for general-purpose crawlers and search engines. In this paper we describe a new hypertext resource discovery system called a Focused Crawler. The goal of a focused crawler is to selectively seek out pages that are relevant to a pre-defined set of topics. The topics are specified not using keywords, but using exemplary documents. Rather than collecting and indexing all accessible Web documents to be able to answer all possible ad-hoc queries, a focused crawler analyzes its crawl boundary to find the links that are likely to be most relevant for the crawl, and avoids irrelevant regions of the Web. This leads to significant savings in hardware and network resources, and helps keep the crawl more up-to-date. To achieve such goal-directed crawling, we designed two hypertext mining programs that guide our crawler: a classifier that evaluates the relevance of a hypertext document with respect to the focus topics, and a distiller that identifies hypertext nodes that are great access points to many relevant pages within a few links. We report on extensive focused-crawling experiments using several topics at different levels of specificity. Focused crawling acquires relevant pages steadily while standard crawling quickly loses its way, even though they are started from the same root set. Focused crawling is robust against large perturbations in the starting set of URLs. It discovers largely overlapping sets of resources in spite of these perturbations. It is also capable of exploring out and discovering valuable resources that are dozens of links away from the start set, while carefully pruning the millions of pages that may lie within this same radius. Our anecdotes suggest that focused crawling is very effective for building high-quality collections of Web documents on specific topics, using modest desktop hardware.

This paper focuses on the representation and access of Web-based information, and how to make such a representation adapt to the activities or interests of individuals within a community of users. The heterogeneous mix of information on the Web restricts the coverage of traditional indexing techniques and so limits the power of search engines. In contrast to traditional methods, and in a way that extends collavotaive filtering approaches, the path model centers representation on usage histories rather than content analysis. By putting activity at the center of representation and not the periphery, the path model concentrates on the reader not the author and the brower not the site. We describe metrics of similarity based on the path model, and their application in a URL recommender tool and in visualising sets of URLs.

The Alexandria Digital Earth Prototype (ADEPT) Project (1999-2004) builds upon the Alexandria Digital Library Project (1994-99) to add functions and services for undergraduate teaching to a digital library of geospatial resources. The Digital Learning Environment (DLE) services are being developed and evaluated iteratively over the course of this research project. In the 2002-2003 academic year, the DLE was implemented in stages during the fall and spring terms in undergraduate geography courses at the University of California, Santa Barbara (UCSB). Evaluation of the fall term implementation identified design issues of time and complexity of use in the services for creating and organizing course domain knowledge. By the time of the spring term implementation, these issues were addressed and new services added for integrating selected course content into a variety of class presentation formats. The implementation was evaluated via interviews with the course instructor, development staff, and students, and by observations (in person and videotaped) of the course. Results of the iterative evaluation indicated that usability and functionality for the instructor had increased between the two course offerings. Students found classroom presentations to be useful for understanding concepts, and Web access to the presentations useful for study and review. Assessments of student learning suggest modest improvements over time. Developers are now applying lessons learned during these implementations to improve the system for subsequent implementation in the 2003-04 academic year.

The combined benefits of smart card to support mobility in a pocket coupled with the ubiquitous access of Web technology, present a new paradigm for medical information access systems. The paper describes the framework of Java Card Web Servlet (JCWS) that is being developed to provide seamless access interface between a Web browser and a Java-enabled smart card. Importantly, the smart card is viewed as a mobile repository of Web objects comprised of HTML pages, medical data objects, and record browsing and updating applet. As the patient moves between hospitals, clinics and countries, the mobility of the smart-card database dynamically binds to the JCWS framework to facilitate a truly ubiquitous access and updating of medical information via a standard Web-browser interface.

Very technical, formal description of query translation. But has the architecture picture.

Extend version of the paper of the same title appeared in TKDE Aug. 1996

Searching over heterogeneous information sources is difficult in part because of the nonuniform query languages. Our approach is to allow users to compose Boolean queries in one rich front-end language. For each user query and target source, we transform the user query into a subsuming query that can be supported by the source but that may return extra documents. The results are then processed by a filter query to yield the correct final results. In this article we introduce the architecture and associated mechanism for query translation. In particular, we discuss techniques for rewriting predicates in Boolean queries into native subsuming forms, which is a basis of translating complex queries. In addition, we present experimental results for evaluating the cost of postfiltering. We also discuss the drawbacks of this approach and cases when it may not be effective. We have implemented prototype versions of these mechanisms and demonstrated them on heterogeneous Boolean systems.

This paper describes G-Portal, a geospatial digital library of geographical assets, providing an interactive platform to engage students in active manipulation and analysis of information resources and collaborative learning activities. Using a G-Portal application in which students conducted a field study of an environmental problem of beach erosion and sea level rise, we described a pilot study to evaluate usefulness and usability issues in supporting geography learning, and in turn teaching.

An asynchronous transfer mode (ATM) wireless network has bursty and high error rates. To combat the contiguous bit loss due to damaged or dropped packets, this paper presents a code packetization and image reconstruction scheme. The packetization method distributes the loss in both frequency and spatial domains to reduce the chance that adjacent DCT blocks lose the same frequency components. The image reconstruction takes into consideration the spatial characteristics represented by the frequency components. Combining these two approaches is able to reconstruct the damaged images more accurately, even under very high loss rates. In addition, since the reconstruction technique is computational efficient, it conserves system resources and power consumption, which are restrictive in mobile computers.

A multimedia server delivers presentations (e.g., videos, movies, providing high bandwidth and continuous real-time deliveryIn this paper we present techniques for reducing the initial latency of presentations, i.e., for reducing the time between the arrival of a request and the start of the presentation. Traditionally, initial latency has not received much attention. This is because one major application of multimedia servers is movies on demand where a delay of a few minutes before a new multi-hour movie starts is acceptable. However , latency reduction is important in interactive applications such as video games and browsing of multimedia documents. V arious latency reduction schemes are proposed and analyzed, and their performance compared. We show that our techniques can signicantly reduce (almost eliminate in some cases) initial latency without adversely affecting throughput. Moreover , a novel on-disk partial data replication scheme that we propose proves to be far more cost effective than any other previous attempts at reducing initial latency. Keywords: multimedia, data placement, data replication.

Collection understanding shifts the traditional focus of retrieval in large collections from locating specific artifacts to gaining a comprehensive view of the collection. Visualization tools are critical to the process of efficient collection understanding. By presenting simple visual interfaces and intuitive methods of interacting with a collection, users come to understand the essence of the collection by focusing on the artifacts. This paper discusses a practical approach for enhancing collection understanding in image collections.

Format: HTML Document (27K + pictures).

Audience: Web surfers and computer scientists. References: 11. Links: 0. Relevance: Low. Abstract: Describes a browser which prefetches pages, and builds a graph showing the relationships of those pages, allows you to jump down in the hierarchy. User specified cutoff for how many nodes should be expanded. No atte

mpt to automatically cluster. Describes naive data structures to implement a breadth first search of the space.

Format: HTML Document (6K) . Audience: Non-technical, standards committee membets. References: 0. Links: 1. Relevance: Medium-low. Abstract: Argues for an IETF rather than ISO model of standards committee. Encouraging several different protocols with gateways being constructed as needed, and generally letting the marketplace determine what survives.

Searching for useful information on the World Wide Web has become increasingly difficult. While Internet search engines have been helping people to search on the web, low recall rate and outdated indexes have become more and more problematic as the web grows. In addition, search tools usually present to the user only a list of search results, failing to provide further personalized analysis which could help users identify useful information and comprehend these results. To alleviate these problems, we propose a client-based architecture that incorporates noun phrasing and self-organizing map techniques. Two systems, namely CI Spider and Meta Spider, have been built based on this architecure. User evaluation studies have been conducted and the findings suggest that the proposed architecture can effectively facilitate web search and analysis.

As the Web has been growing exponentially, it has become increasingly difficult to search for desired information. In recent years, many domain-specific (vertical) search tools have been developed to serve the information needs of specific fields. This paper describes two approaches to building a domain-specific search tool. We report our experience in building two different tools in the nanotechnology domain - (1) a server-side search engine, and (2) a client-side search agent. The designs of the two search systems are presented and discussed, and their strengths and weaknesses are compared. To our knowledge, this paper is the first to compare prototype vertical search systems built by the two different approaches. Some future research directions are also discussed.

In this paper we discuss the architecture of a tool designed to help users develop vertical search engines in different domains and different languages. The design of the tool is presented and an evaluation study was conducted, showing that the system is easier to use than other existing tools.

Format: HTML Document (8K) .

Audience: Researchers, esp. in the area of multi-media searching. References: 8. Links: 1. Relevance: Low-Medium. Abstract: Essentially a set of pointers to Xerox PARC reports. Describes projects related to scatter/gather, automatic segmenting, keyword search equivalents for audio & video, and summarization.

An Archival Intermemory solves the problem of highly survivable digital data storage in the spirit of the Internet. In this paper we describe a prototype implementation of Intermemory, including an overall system architecture and implementations of key system components. The result is a working Intermemory that tolerates up to 17 simultaneous node failures, and includes a Web gateway for browser-based access to data. Our work demonstrates the basic feasibility of Intermemory and represents significant progress towards a deployable system.

Format: HTML Document().

In Web-based searching systems that access distributed information providers, efficient query processing requires an advanced caching mechanism to reduce the query response time. The keyword-based querying is often the only way to retrieve data from Web providers, and therefore standard page-based and tuple- based caching mechanisms turn out to be improper for such a task. In this work, we develop a mechanism for efficient caching of Web queries and the answers received from heterogeneous Web providers. We also report results of experiments and show how the caching mechanism is implemented in the Knowledge Broker system.

XML Schema language has been proposed to replace Document Type Definitions (DTDs) as schema mechanism for XML data. This language consistently extends grammar-based constructions with constraint- and pattern-based ones and have a higher expressive power than DTDs. As schemas remain optional for XML, we address the problem of XML Schema extraction. We model the XML schema as extended context-free grammars and develop a novel extraction algorithm inspired by methods of grammatical inference. The algorithm copes also with the schema determinism requirement imposed by XML DTDs and XML Schema languages.

In this paper we study how to build an effective incremental crawler. The crawler selectively and incrementally updates its index and/or local collection of web pages, instead of periodically refreshing the collection in batch mode. The incremental crawler can improve the ``freshness'' of the collection significantly and bring in new pages in a more timely manner. We first present results from an experiment conducted on more than half million web pages over 4 months, to estimate how web pages evolve over time. Based on these experimental results, we compare various design choices for an incremental crawler and discuss their trade-offs. We propose an architecture for the incremental crawler, which combines the best design choices.

In this paper we study how to refresh a local copy of an autonomous data source to maintain the copy up-to-date. As the size of the data grows, it becomes more difficult to maintain the copy fresh, making it crucial to synchronize the copy effectively. We define two freshness metrics, change models of the underlying data, and synchronization policies. We analytically study how effective the various policies are. We also experimentally verify our analysis, based on data collected from 270 web sites for more than 4 months, and we show that our new policy improves the freshness very significantly compared to current policies in use.

In this paper we study in what order a crawler should visit the URLs it has seen, in order to obtain more important pages first. Obtaining important pages rapidly can be very useful when a crawler cannot visit the entire Web in a reasonable amount of time. We define several importance metrics, ordering schemes, and performance evaluation measures for this problem. We also experimentally evaluate the ordering schemes on the Stanford University Web. Our results show that a crawler with a good ordering scheme can obtain important pages significantly faster than one without.

They crawl the Web and automatically find out which sites completely or partially mirror each other.

Many web documents (such as JAVA FAQs) are being replicated on the Internet. Often entire document collections (such as hyperlinked Linux manuals) are being replicated many times. In this paper, we make the case for identifying replicated documents and collections to improve web crawlers, archivers, and ranking functions used in search engines. The paper describes how to efficiently identify replicated documents and hyperlinked document collections. The challenge is to identify these replicas from an input data set of several tens of millions of web pages and several hundreds of gigabytes of textual data. We also present two real-life case studies where we used replication information to improve a crawler and a search engine. We report these results for a data set of 25 million web pages (about 150 gigabytes of HTML data) crawled from the web

While it would seem that digital video libraries should benefit from access mechanisms directed to their visual contents, years of TREC Video Retrieval Evaluation (TRECVID) research have shown that text search against transcript narrative text provides almost all the retrieval capability, even with visually oriented generic topics. A within-subjects study involving 24 novice participants on TRECVID 2004 tasks again confirms this result. The study shows that satisfaction is greater and performance is significantly better on specific and generic information retrieval tasks from news broadcasts when transcripts are available for search. Additional runs with 7 expert users reveal different novice and expert interaction patterns with the video library interface, helping explain the novices’ lack of success with image search and visual feature browsing for visual information needs. Analysis of TRECVID visual features well suited for particular generic tasks provides additional insights into the role of automated feature classification for digital image and video libraries.

Surrogates, summaries, and visualizations have been developed and evaluated for accessing a digital video library containing thousands of documents and terabytes of data. These interfaces, formerly implemented within a monolithic stand-alone application, are being migrated to XML and XSLT for delivery through web browsers. The merits of these interfaces are presented, along with a discussion of the benefits in using W3C recommendations such as XML and XSLT for delivering tailored access to video over the web.

For many readers, handling a physical book is an enjoyably exquisite part of the information seeking process. Many physical characteristics of a book its size, heft, the patina of use on its pages and so on communicate ambient qualities of the document it represents. In contrast, the experience of accessing and exploring digital library documents is dull. The emphasis is utilitarian; technophile rather than bibliophile. We have extended the page-turning algorithm we reported at last year's JCDL into a scaleable, systematic approach that allows users to view and interact with realistic visualizations of any textual-based document in a Greenstone collection. Here, we further motivate the approach, illustrate the system in use, discuss the system architecture and present a user evaluation. Our work leads us to believe that far from being a whimsical gimmick, physical book models can usefully complement conventional document viewers and increase the perceived value of a digital library system.

Can digital libraries provide a reading experience that more closely resembles a real book than a scrolled or paginated electronic display? This paper describes a prototype page-turning system that realistically animates full three-dimensional page-turns. The dynamic behavior is generated by a mass-spring model defined on a rectangular grid of particles. The prototype takes a PDF or E-book file, renders it into a sequence of PNG images representing individual pages, and animates the pageturns under user control. The simulation behaves fairly naturally, although more computer graphics work is required to perfect it.

We describe a Web-based application developed above a digital library of geographical resources for Singapore students preparing to take a national examination in geography. The application provides an interactive, non-sequential approach to learning that supplements textbooks.

As part of the Illinois Digital Library Initiative (DLI) project we developed scalable semantics technologies. These statistical techniques enabled us to index large collections for deeper search than word matching. Through the auspices of the DARPA Information Management program, we are developing an integrated analysis environment, the Interspace Prototype, that uses semantic indexing as the foundation for supporting concept navigation. These semantic indexes record the contextual correlation of noun phrases, and are computed generically, independent of subject domain.

Using this technology, we were able to compute semantic indexes for a subject discipline. In particular, in the summer of 1998, we computed concept spaces for 9.3M MEDLINE bibliographic records from the National Library of Medicine (NLM) which extensively covered the biomedical literature for the period from 1966 to 1997. In this experiment, we first partitioned the collection into smaller collections (repositories) by subject, extracted noun phrases from titles and abstracts, then performed semantic indexing on these subcollections by creating a concept space for each repository. The computation required 2 days on a 128-node SGI/CRAY Origin 2000 at the National Center for Supercomputer Ap- plications (NCSA). This experiment demonstrated the feasibility of scalable semantics techniques for large collections. With the rapid increase in computing power, we believe this indexing technology will shortly be feasible on personal computers.

We present an ongoing project to develop a system to provide field service workers with timely and accurate service information. The system will allow workers to download diagrams or photographs from a host computer's central database onto a PDA. The workers will be able to annotate the diagrams to reflect work performed, and later upload the annotations to the host computer, where they will be integrated into an updated database. Diagram recognition functionality is distributed between the PDA (which performs low-level shape and handwriting recognition) and the host computer (which performs high-level domain-based diagram recognition). Distributing the functionality offers a number of advantages: it allows the relatively resource-poor PDA to be part of a powerful diagram recognition environment, it allows the use of standardized hardware-based recognition facilities in a domain-based recognition system, and it allows off-line drawing recognition and storage of diagrams, thereby avoiding excessive use of slow or expensive communications channels.

Hyper Text Transfer Protocol (HTTP) traffic dominates Internet traffic. The exchange of HTTP messages is implemented using the connection-oriented TCP. HTTP/l.0 establishes a new TCP connection for each HTTP request, resulting in many consecutive short-lived TCP connections. The emerging HTTP/ 1.1 reduces latencies and overhead from closing and re-establishing connections by supporting persistent connections as a default. A TCP connection which is kept open and reused for the next HTTP request reduces overhead and latency. Open connections, however, consume sockets and memory for socket-buffers. This trade-off establishes a need for connection-management policies. We propose policies that exploit embedded information in the HTTP request messages, e.g., senders' identities and requested URLs. and compare them to the fixed-timeout policy used in the current implementation of the Apache Web server. An experimental evaluation of connection management policies at Web servers, conducted using Web server logs. shows that our URL-based policy consistently outperforms other policies. and achieves significant 15-25with respect to the fixed-timeout policy. Hence, allowing Web servers and clients to more fully reap the benefits of persistent HTTP.

One bottleneck in implementing a system that intelligently queries the Web is developing `wrappers' - programs that extract data from Web pages. Here we describe a method for learning general, page-independent heuristics for extracting data from HTML documents. The input to our learning system is a set of working wrapper programs, paired with HTML pages they correctly wrap. The output is a general procedure for extracting data that works for many formats and many pages. In experiments with a collection of 84 constrained but realistic extraction problems, we demonstrate that 30problems can be handled perfectly by learned extraction heuristics, and around 50handled acceptably. We also demonstrate that learned page-independent extraction heuristics can substantially improve the performance of methods for learning page-specific wrappers.

We describe an image retrieval system we built based on a Zoomable User Interface (ZUI). We also discuss the design, results and analysis of a controlled experiment we performed on the browsing aspects of the system. The experiment resulted in a statistically significant difference in the interaction between number of images (25, 75, 225) and style of browser (2D, ZUI, 3D). The 2D and ZUI browser systems performed equally, and both performed better than the 3D systems. The image browsers tested during the experiment include Cerious Software's Thumbs Plus, TriVista Technology's Simple Landscape and Photo GoRound, and our Zoomable Image Browser based on Pad++.

Format: HTML Document().

Digital materials can be protected from failures by replicating them at multiple autonomous, distributed sites. A significant challenge in such a distributed system is ensuring that documents are replicated and accessible despite malicious sites. Such sites may hinder the replication of documents in a variety of ways, including agreeing to store a copy but erasing it instead, refusing to serve a document, or serving an altered version of the document. We describe the design of a a Peer-to-peer Information Preservation and Exchange (PIPE) network: a distributed replication system that protects documents both from failures and from malicious nodes. We present the design of a PIPE system, discuss a threat model for malicious sites, and propose basic solutions for managing these malicious sites.

Digital materials can be protected from failures by replicating them at multiple autonomous, distributed sites. A Peer-to-peer Information Preservation and Exchange (PIPE) network is a good way to build a distributed replication system. A significant challenge in such networks is ensuring that documents are replicated and accessible despite malicious sites. Such sites may hinder the replication of documents in a variety of ways, including agreeing to store a copy but erasing it instead, refusing to serve a document, or serving an altered version of the document. We define a model of PIPE networks, a threat model for malicious sites, and propose basic solutions for managing these malicious sites. The basic solutions are inefficient, but demonstrate that a secure system can be built. We also sketch ways to improve the efficiency of the system.

An Archival Repository reliably stores digital objects for long periods of time (decades or centuries). The archival nature of the system requires new techniques for storing, indexing, and replicating digital objects. In this paper we discuss the specialized indexing needs of a write-once archive. We also present a reliability algorithm for effectively replicating sets of related objects. We describe an administrative user interface and a data import utility for archival repositories. Finally, we discuss and evaluate a prototype repository we have built, the Stanford Archival Vault, SAV.

It may be important to provide long-term preservation of digital data even when that data is stored in an unreliable system, such as a filesystem, a legacy database, or even the World Wide Web. In this research paper we focus on the problem of archiving the contents of a web site without disrupting users who maintain the site. We propose an archival storage system, the InfoMonitor, in which a reliable archive is integrated with an unmodified existing store. Implementing such a system presents various challenges related to the mismatch of features between the components, such as differences in naming and data manipulation operations. We examine each of these issues as well as solutions for the conflicts that arise. We also discuss our experience using the InfoMonitor to archive the Stanford Database Group's web site.

Data archiving systems rely on replication to preserve information. In this paper, we discuss how a network of autonomous archiving sites can trade data to achieve the most reliable replication. A series of binary trades between sites produces a peer to peer archiving network. We examine two trading algorithms, one based on trading collections (even if they are different sizes) and another based on trading equal sized blocks of space (which can then store collections.) We introduce the concept of deeds, which track the sites that own space at other sites. We then discuss policies for tuning these algorithms to provide the highest reliability, for example by changing the order in which sites are contacted and offered trades. Finally, we present simulation results that reveal which policies are most reliable.

Data archiving systems rely on replication to preserve information. In this paper, we discuss how a network of autonomous archiving sites can trade data to achieve the most reliable replication. A series of binary trades between sites produces a peer to peer archiving network. We examine two trading algorithms, one based on trading collections (even if they are different sizes) and another based on trading equal sized blocks of space (which can then store collections.) We introduce the concept of deeds, which track the sites that own space at other sites. We then discuss policies for tuning these algorithms to provide the highest reliability, for example by changing the order in which sites are contacted and offered trades. Finally, we present simulation results that reveal which policies are most reliable.

Digital archives protect important data collections from failures by making multiple copies at other archives, so that there are always several good copies of a collection. In a cooperative replication network, sites ``trade'' space, so that each site contributes storage resources to the system and uses storage resources at other sites. Here, we examine bid trading: a mechanism where sites conduct auctions to determine who to trade with. A local site wishing to make a copy of a collection announces how much remote space is needed, and accepts bids for how much of its own space the local site must ``pay'' to acquire that remote space. We examine the best policies for determining when to call auctions and how much to bid, as well as the effects of ``maverick'' sites that attempt to subvert the bidding system. Simulations of auction and trading sessio ns indicate that bid trading can allow sites to achieve higher reliability than the alternative: a system where sites trade equal amounts of space without bidding.

Digital archives can best survive failures if they have made several copies of their collections at remote sites. In this paper, we discuss how autonomous sites can cooperate to provide preservation by trading data. We examine the decisions that an archive must make when forming trading networks, such as the amount of storage space to provide and the best number of partner sites. We also deal with the fact that some sites may be more reliable than others. Experimental results from a data trading simulator illustrate which policies are most reliable. Our techniques focus on preserving the ``bits'' of digital collections; other services that focus on other archiving concerns (such as preserving meaningful metadata) can be built on top of the system we describe here.

Digital archives can best survive failures if they have made several copies of their collections at remote sites. In this paper, we discuss how autonomous sites can cooperate to provide preservation by trading data. We examine the decisions that an archive must make when forming trading networks, such as the amount of storage space to provide and the best number of partner sites. We also deal with the fact that some sites may be more reliable than others. Experimental results from a data trading simulator illustrate which policies are most reliable.

Digital archives can best survive failures if they have made several copies of their collections at remote sites. In this paper, we discuss how autonomous sites can cooperate to provide preservation by trading data. We examine the decisions that an archive must make when forming trading networks, such as the amount of storage space to provide and the best number of partner sites. We also deal with the fact that some sites may be more reliable than others. Experimental results from a data trading simulator illustrate which policies are most reliable. Our techniques focus on preserving the ``bits'' of digital collections; other services that focus on other archiving concerns (such as preserving meaningful metadata) can be built on top of the system we describe here.

Data archiving systems rely on replication to preserve information. In this paper, we discuss how a network of autonomous archiving sites can trade data to achieve the most reliable replication. A series of binary trades between sites produces a peer to peer archiving network. We examine two trading algorithms, one based on trading collections (even if they are different sizes) and another based on trading equal sized blocks of space (which can then store collections.) We introduce the concept of deeds, which track the sites that own space at other sites. We then discuss policies for tuning these algorithms to provide the highest reliability, for example by changing the order in which sites are contacted and offered trades. Finally, we present simulation results that reveal which policies are most reliable.

Data archiving systems rely on replication to preserve information. In this paper, we discuss how a network of autonomous archiving sites can trade data to achieve the most reliable replication. A series of binary trades between sites produces a peer to peer archiving network. We examine two trading algorithms, one based on trading collections (even if they are different sizes) and another based on trading equal sized blocks of space (which can then store collections.) We introduce the concept of deeds, which track the sites that own space at other sites. We then discuss policies for tuning these algorithms to provide the highest reliability, for example by changing the order in which sites are contacted and offered trades. Finally, we present simulation results that reveal which policies are most reliable.

Digital archives protect important data collections from failures by making multiple copies at other archives, so that there are always several good copies of a collection. In a cooperative replication network, sites ``trade'' space, so that each site contributes storage resources to the system and uses storage resources at other sites. Here, we examine bid trading: a mechanism where sites conduct auctions to determine who to trade with. A local site wishing to make a copy of a collection announces how much remote space is needed, and accepts bids for how much of its own space the local site must ``pay'' to acquire that remote space. We examine the best policies for determining when to call auctions and how much to bid, as well as the effects of ``maverick'' sites that attempt to subvert the bidding system. Simulations of auction and trading sessio ns indicate that bid trading can allow sites to achieve higher reliability than the alternative: a system where sites trade equal amounts of space without bidding.

The popularity of peer-to-peer search networks grows, even as the limitations to the scalability of existing systems become apparent. We propose a simple model for search networks, called the search/index links (SIL) model. The SIL model describes existing networks while also yielding organizations not previously studied. Using simulation results, we argue that a new organization, parallel search clusters, is superior to existing supernode networks in many cases.

The popularity of peer-to-peer search networks grows, even as the limitations to scalability of existing systems becomes apparent. We propose a simple model for search networks, called the search/index links (SIL) model. The SIL model describes existing networks while also yielding organizations not previously studied. Using simulation results, we argue that a new organization, parallel search clusters, is superior to existing supernode networks in many cases.

We have applied speech recognition and text-mining technologies to a set of recorded outbound marketing calls and analyzed the results. Since speaker-independent speech recognition technology results in a significantly lower recognition rate than that found when the recognizer is trained for a particular speaker, we applied a number of post-processing algorithms to the output of the recognizer to render it suitable for the Textract text mining system. We indexed the call transcripts using a search engine and used Textract and associated Java technologies to place the relevant terms for each document in a relational database. Following a search query, we generated a thumbnail display of the results of each call with the salient terms highlighted. We illustrate these results and discuss their utility. We took the results of these experiments and continued this analysis on a set of talks and presentations. We describe a distinct document genre based on the note-taking concept of document content, and propose a significant new method for measuring speech recognition accuracy. This procedure is generally relevant to the problem of capturing meetings and talks and providing a searchable index of these presentations on the web.

In this paper, we describe the JAFER ToolKit project which is developing a simplified XML based API above the Z39.50 protocol. The ToolKit allows the development of both Z39.50 based applications (both clients and servers) without detailed knowledge of the complexities of the protocol.

Steve Cousin's Ph.D. thesis

Describes how the DLITE design enables shifting of functionality among distributed components.

Format: HTML Document(39K + pictures) . Audience: Computer Scientists. References: 10. Links: 8. Relevance: High. Abstract: Describes an architecture called InterPay for allowing heterogeneous payment mechanisms to interoperate. Defines three levels (a task level, payment policy level, and payment mechanism level) that may be modified in

dependently. Describes a working prototype using the ILU distributed object system from Xerox. Shows a sample transaction using the architecture, and how the components of the architecture (payment agents, collection agents, and paym

ent and collection capabilities) can be used in more complex transactions.

Interpay paper

We have designed and prototyped a series of interfaces for Digital Libraries. These interfaces use CORBA objects to distribute interface modeling and rendering across machines. We describe the design tensions arising in the context of such distribution, locate existing UI technology in the resulting design space, and explain the location of our final prototype in that space. We view Digital Libraries as collections of repositories and publication-related services that may be distributed over large distances and must be accessible from many locations and through multiple hardware, software, and networking platforms. We describe our use of CORBA and briefly introduce a drag-and-drop interface developed to provide unified access to heterogeneous Digital Library resources.

Format: Not yet online.

This paper describes the creation of a new humanities digital library collection: 11,000,000 words and 10,000 images representing books, images and maps on pre-twentieth century London and its environs. The London collection contained far more dense and precise information than the materials from the Grecco-Roman world on which we had previously concentrated. The London collection thus allowed us to explore new problems of data structure, manipulation, and visualization. This paper contrasts our model for how humanities digital libraries are best used with the assumptions that underlie many academic digital libraries on the one hand and more literary hypertexts on the other. Since encoding guidelines such as those from the TEI provide collection designers with far more options than any one project can realize, this paper describes what structures we used to organize the collection and why. We particularly emphasize the importance of mining historical authority lists (encyclopedias, gazetteers, etc.) and then generating automatic span-to-span links within the collection.

This paper surveys research areas relevant to cultural heritage digital libraries. The emerging National Science Digital Library promises to establish the foundation on which those of us beyond the scientific and engineering community will likely build. This paper thus articulates the particular issues that we have encountered in developing cultural heritage collections. We provide a broad overview of audiences, collections, and services.

We evaluate server selection methods in a Web environment, modeling a digital library which makes use of existing Web search servers rather than building its own index. The evaluation framework portrays the Web realistically in several ways. Its search servers index real Web documents, are of various sizes, cover different topic areas and employ different retrieval methods. Selection is based on statistics extracted from the results of probe queries submitted to each server. We evaluate published selection methods and a new method for enhancing selection based on expected search server effectiveness. Results show CORI to be the most effective of three published selection methods.

This paper introduces techniques for reducing data dissemination costs of query subscriptions. The reduction is achieved by merging queries with overlapping, but not necessarily equal, answers. The paper formalizes the query-merging problem and introduces a general cost model for it. We prove that the problem is NP-hard and propose exhaustive algorithms and three heuristic algorithms: the Pair Merging Algorithm, the Directed Search Algorithm and the Clustering Algorithm. We develop a simulator for evaluating the different heuristics and show that the performance of our heuristics is close to optimal.

Traditional server-based web applications allow access to server-hosted resources, but often exhibit poor responsiveness due to server load and network delays. Client-side web applications, on the other hand, provide excellent interactivity at the expense of limited access to server resources. The WebWriter II Editor, a direct manipulation HTML editor that runs in a web browser, uses both server-side and client-side processing in order to achieve the advantages of both. In particular, this editor downloads the document data structure to the browser and performs all operations locally. The user interface is based on HTML frames and includes individual frames for previewing the document and displaying general and specific control panels. All editing is done by JavaScript code residing in roughly twenty HTML pages that are downloaded into these frames as needed. Such a client-server architecture, based on frames, client-side data structures, and multiple JavaScript-enhanced HTML pages appears promising for a wide variety of applications. This paper describes this architecture, the Meteor Shower Application Architecture, and its use in the WebWriter II Editor.

We propose an architecture for Digital Library Repositories that assures long-term archival storage of digital objects. The architecture is formed by a federation of independent but collaborating sites, each managing a collection of digital objects. The architecture is based on the following key components: use of signatures as object handles, no deletions of digital objects, functional layering of services, the presence of an awareness service in all layers, and use of disposable auxiliary structures. Long-term persistence of digital objects is achieved by creating replicas at several sites.

This paper studies the archival problem: how a digital library can preserve electronic documents over long periods of time. We analyze how an archival repository can fail and we present different strategies that help solve the problem. We introduce ArchSim, a simulation tool that for evaluating an implementation of an archival repository system and compare options such as different disk reliabilities, error detection and correction algorithms, preventive maintenance, etc. We use ArchSim to analyze a case study of an Archival Repository for Computer Science Technical Reports.

Designing an archival repository is a complex task because there are many alternative configurations, each with different reliability levels and costs. In this paper we study the costs involved in an Archival Repository and we introduce a design framework for evaluating alternatives and choosing the best configuration in terms of reliability and cost. We also present a new version of our simulation took, ArchSim/C that aids in the decision process. The design framework and the usage of ArchSim/C are illustrated with a case study of a hypothetical (yet realistic) archival repository shared between two universities.

Finding information in a peer-to-peer system currently requires either a costly and vulnerable central index, or flooding the network with queries. In this paper we introduce the concept of Routing Indices (RIs), which allow nodes to forward queries to neighbors that are more likely to have answers. If a node cannot answer a query, it forwards the query to a subset of its neighbors, based on its local RI, rather than by selecting neighbors at random or by flooding the network by forwarding the query to all neighbors. We present three RI schemes: the compound, the hop-count, and the exponential routing indices. We evaluate their performance via simulations, and find that RIs can improve performance by one or two orders of magnitude vs. a flooding-based system, and by up to 100the different RI schemes and highlight the effects of key design variables on system performance.

This paper presents the design and the current prototype implementation of an interactive vocal Information Retrieval system that can be used to access articles of a large news paper archive using a telephone. The results of preliminary investigation into the feasibility of such a system are also presented.

Format: HTML Document (39K + picture). . Audience: Computer scientists and librarians References: 10. Links: 1. Relevance: Low. Abstract: Argues that many of the problems in DL development are not technical, but social and political, as the nature of the work is transformed. Describes two Xerox collaborations with academia, one on scanned documents, a

second with a web-based system. Lists some assumptions that are sometimes made, and to what extent they are borne out. Special concerns are standards, which ones to support, and ensuring access to acquisitions in old standards.

Format: HTML Document().

Format: HTML Document(29K) .

Audience: Information retrieval specialists. References: 12. Links: 1. Relevance: Low. Abstract: Describes use of THOMAS, the on-line source of congressional information. Based on the INQUERY engine, offers keyword searches with other advanced features (proximity, weighted averaging, synonyms) which are largely ignored by the user population. The tendency is for short (3 word or less) queries about a single topic. Describes domain-dependent performance enhancements with the ranking algorithms to ensure that relevant hits appear near the top

of the ranking.

Format: PostScript () .

Browsing through collections and reading activities are separated in most present WWW-based user's interfaces of digitalized libraries. This context break induces longer apprenticeship and navigation time within the interface. We study in this paper how 3D interaction metaphors can be used to provide a continuous navigation space for these two tasks.

In this paper, we explore an approach to make better use of semi-structured documents in information retrieval in the domain of biology. Using machine learning techniques, we make those inherent structures explicit by XML markups. This marking up has great potentials in improving task performance in specimen identification and the usability of online flora and fauna.

When people are looking for visual arts information - information related to images - how do they characterize their needs? We analyze a set of 405 queries to identify the attributes that people provide to the Google Answers' ask an expert online reference system. The results suggest directions to take in developing an effective organization and features for an image digital library.

In this paper we describe the results of an ethnographic study of the information behaviours of university technical support workers and their information needs. The study looked at how the group identified, located and used information from a variety of sources to solve problems arising in the course of their work. The results of the investigation are discussed in the context of the feasibility of developing a potential information base that could be used by all members of the group. Whilst a number of their requirements would easily be fulfilled by the use of a digital library, other requirements would not. The paper illustrates the limitations of a digital library with respect to the information behaviours of this group of subjects and focuses on why a digital library would not appear to be the ideal support tool for their work.

At present, music digital library systems are being developed based on anecdotal evidence of user needs, intuitive feelings for user information seeking behavior, and a priori assumptions of typical usage scenarios. Emphasis has instead been placed on basic research into music document representation, efficient searching, and audio-based searching, rather than on exploring the music information needs or information behavior of a target user group. This paper focuses on eliciting the 'native' music information strategies employed by people searching for popular music (that is, music sought for recreational or enjoyment purposes rather than to support a 'serious' or scientific exploration of some aspect of music). To this end, we conducted an ethnographic study of the searching/browsing techniques employed by people in the researchers' local communities, as they use two common sources of music: the public library and music stores. We argue that the insights provided by this type of study can inform the development of searching/browsing support for music digital libraries.

This paper examines user choice of interface language in a bi-language digital library (English and Māori, the language of the indigenous people of New Zealand). The majority of collection documents are in Māori, and the interface is available in both Māori and English. Log analysis shows three categories of preference for interface language: primarily English, primarily Māori, and bilingual (switching back and forth between the two). As digital libraries increase in number, content, and potential user base, interest has grown in ‘multilingual’ or ‘multi-language’ collections-that is, digital libraries in which the collection documents and the collection interface include more than one language. Research in multilingual/multi-language digital libraries and web-based document collections has primarily focused on fundamental implementation issues and functionality, principles for design, and small-scale usability tests; at present no analysis exists of how these systems are used, or how the presence of more than one language in a digital library affects user interactions-presumably because multilingual/multi-language digital libraries are only recently moving from research lab prototypes to fielded systems, and few have built up a significant usage history. This paper describes the application of log analysis to examine interface language preference in a bi-language (English/Māori) digital library-the Niupepa Collection (Section 2). Web log data was collected for a year (Section 3), and log analysis indicates three categories of interface language preferences: English, Māori, and ‘bilingual’ (Section 4). A fine-grained analysis of activities within user sessions indicates different patterns of document access and information gathering strategy between these three categories (Section 5).

Reference manual. Tech report at cour94

The Digital Anthropology Resources for Teaching (DART) project integrates the content acquisition and cataloging initiatives of a federated digital repository with the development of scholarly publications and the creation of digital tools to facilitate classroom teaching. The project's technical architecture and unique publishing model create a teaching context where students move easily between primary and secondary source material and between authored environments and independent research, and raise specific issues with regard to metadata, object referral, rights, and exporting content. The model also addresses the loss of provenance and catalog information for digital objects embedded in born-digital publications. The DART project presents a practical methodology to combine repository and publication that is both exportable and discipline-neutral.

Distributed collections of Web materials are common. Bookmark lists, paths, and catalogs such as Yahoo! Directories require human maintenance to keep up to date with changes to the underlying documents. The Walden's Paths Path Manager is a tool to support the maintenance of distributed collections. Earlier efforts focused on recognizing the type and degree of change within Web pages and identifying pages no longer accessible. We now extend this work with algorithms for evaluating drastic changes to page content based on context. Additionally, we expand on previous work to locate moved pages and apply the modified approach to suggesting page replacements when the original page cannot be found. Based on these results we are redesigning the Path Manager to better support the range of assessments necessary to manage distributed collections.

Distributed collections of Web materials are common. Bookmark lists, paths, and catalogs such as Yahoo! Directories require human maintenance to keep up to date with changes to the underlying documents. The Walden's Paths Path Manager is a tool to support the maintenance of distributed collections. Earlier efforts focused on recognizing the type and degree of change within Web pages and identifying pages no longer accessible. We now extend this work with algorithms for evaluating drastic changes to page content based on context. Additionally, we expand on previous work to locate moved pages and apply the modified approach to suggesting page replacements when the original page cannot be found Based on these results we are redesigning the Path Manager to better support the range of assessments necessary to manage distributed collections.

Format: Not yet online.

Format: Not Yet Online.

Publishers wishing to distribute text online fear that customers will download their product and redistribute it illegally. Although constraining the users to access the data only through proprietary software that does not allow downloading helps, it still leaves the possibility that users could take screen dumps of the material to capture it. The technique described in the paper relies on the perceptual properties of the human eye, using two unreadable images interleaved quickly to create a readable image, which cannot be screen-dumped since the readability depends on averaging in the human eye. Our program flickers two images of the text each with an admixture of grey noise. Your eye sorts out the letters and reads them, not paying close attention to the grey background; but any screen dump captures the item at one instant including the noise. The text is also scrolled up and down slowly, which again your eye can track, but which would frustrate a program trying to average out the flickering.

In a Peer-To-Peer (P2P) system, autonomous computers pool their resources (e.g., files, storage, compute cycles) in order to inexpensively handle tasks that would normally require large costly servers. The scale of these systems, their open nature, and the lack of centralized control pose difficult performance and security challenges. Much research has recently focused on tackling some of these challenges; in this paper, we propose future directions for research in P2P systems, and highlight problems that have not yet been studied in great depth. We focus on two particular aspects of P2P systems - search and security - and suggest several open and important research problems for the community to address.

Research in Peer-to-peer systems has focussed on building efficient Content Addressable Networks (CANs), which are essentially distributed hash tables (DHT) that support location of resources based on unique keys. While most proposed schemes are robust to a large number of random faults, there are very few schemes that are robust to a large number of adversarial faults. In a recent paper Fiat and Saia have proposed such a solution that is robust to adversarial faults. We propose a new solution based on multi-butterflies that improves upon the previous solution by Fiat and Saia. Our new network, multi-hypercube, is a fault tolerant version of the hypercube, and may find applications to other problems as well. We also demonstrate how this network can be maintained dynamically. This addresses the first open problem in the paper by Fiat and Saia.

The popularity of systems like Napster, Gnutella etc. have spurred recent interest in Peer-to-peer systems. A central problem in all these systems is efficient location of resources based on their keys. A network that supports such queries is referred to as Content Addressable Network (CAN). Many solutions have been proposed to building CANs. However most of these solutions do not focus on adversarial faults, which might be critical to building a censorship resistant peer-to-peer system. In a recent paper Fiat and Saia have proposed a solution to building such a system. We propose a new solution based on multi-butterflies that improves upon the previous solution by Fiat and Saia. Our new network, ( multi-hypercube), is a fault tolerant version of hypercube. We also demonstrate how this network can be maintained dynamically. This addresses the first open problem in the paper by Fiat and Saia.

Format: Compressed PostScript().

Format: HTML Document (6K) .

Audience: UK funders . References: 4. Links: 1. Relevance: Low. Abstract: A description of the Microcosm system (campus document delivery), a hypermedia system allowing links to 3rd party viewers.

Format: HTML Document(34K + pictures) .

Audience: Digitial library developers and users. References: 21. Links: 1. Relevance: Low. Abstract: Rather than just retrieving a list of hits for a query, the system can bundle them, generating hyperlinks on keywords, offer interactive query expansion or contraction. Suggests the addition of a length (in minu

tes to comprehend) and a reader level field of meta-information.

Format: Not Yet Online.

Format: HTML Document().

Format: Not yet online.

In this paper, we describe an interactive system. built within the context of CLiMB project, which permits a user to locate the occurrences of named entities within a given text. The named entity tool was developed to identify references to a single art object (e.g. a particular building) with high precision in text related to images of that object in a digital collection. We start with an authoritative list of art objects, and seek to match variants of these named entities in related text. Our approach is to decay entities into progressively more general variants while retaining high precision. As variants become more general, and thus more ambiguous, we propose methods to disambiguate intermediate results. Our results will be used to select records into which automatically generated metadata will be loaded.

Format: HTML Document (31K) . Audience: Generalist, academic. References: 1. Links: 0. Relevance: low-medium. Abstract: Discusses the 4 services of publisher and library: Gathering, Selecting, Enhancing, and Informing in terms of benefits provided to academics and society. Argues that distribution of ideas is too important to be exclusively at the mercy of the market place, and should (like theater or public TV) be subsidized, but the majority of cost recovery should still be from users. A

rgues that the producers and consumers (university presses and faculty) are largely part of the same institution, so there should be gains, but presses have evolved to be largely independent.

Format: HTML Document (21K) .

Audience: publishers, librarians. References: 0. Links: 0 . Relevance: low-medium . Abstract: Economic discussion of publishing. Looks at first copy and incremental copy costs. Considers ways that publishers can recover first copy costs while still distributing to all for whom it is economically rational (value is greater than incremental cost.) Possible models: 1) country club, where one pays high up-front dues, but then low per-transaction cost; 2) differentiated costs where different products are provided, one at a higher cost with certain features, a second at lower (marginal) cost, e.g., more expensive hardcover comes out first, followed by cheap paperback months later. Mentions 3 specific examples: Project Muse, Chicago Journal of Theoretical Computer Science, and Mathematical Reviews.

Format: Not Yet Online.

Abstract: In this paper, we illustrate the use of space-time trade-offs for information presentation on small screens. We propose the use of Rapid Serial Visual Presentation (RSVP) to provide a rich set of navigational information for Web browsing. The principle of RSVP browsing is applied to the development of a Web browser for small screen devices, the RSVP browser. The results of an experiment in which Web browsing with the RSVP browser is compared with that of a typical WAP browser suggests that RSVP browsing may indeed offer alternative to other forms of Web browsing on small screen devices.

When using traditional search engines, users have to formulate queries to describe their information need. This paper discusses a different approach to Web searching where the input to the search process is not a set of query terms, but instead is the URL of a page, and the output is a set of related Web pages. A related Web page is one that addresses the same topic as the original page. For example, www.washingtonpost.con is a page related to www.nytimes.con, since both are online newspapers. We describe two algorithms to identify related Web pages. These algorithms use only the connectivity information in the Web (i.e., the links between pages) and not the content of pages or usage information. We have implemented both algorithms and measured their runtime performance. To evaluate the effectiveness of our algorithms, we performed a user study comparing our algorithms with Netscape's `What's Related' service (http://home.netscape.con/escapes/related/). Our study showed that the precision at 10 for our two algorithms are 73despite the fact that Netscape uses both content and usage pattern information in addition to connectivity information.

Format: Compressed PostScript().

This paper discusses four kinds of security controls: access control, flow control, inference control, and data encryption. It describes the general nature of controls of each type, the kinds of problems they can and cannot solve, and their inherent limitations and weakness.

We expect that people will want to search for video news or entertainment on mobile platforms as soon as the technology is ready. An Ipaq palmtop version of the Informedia Digital Video Library interface has already been developed at the Chinese University of Hong Kong. Separately, we used the Desktop Informedia interface for the interactive part of the Trec10 video track competition. The lesson we learned is that automated image search is so poor that the best interactive results come from showing the user many images quickly, and allowing flexible drill down to images from nearby shots. Here we report on an effort to apply this lesson to palmtop platforms, where showing a large grid of images in parallel is not feasible. Perceptual psychology experiments suggest that time-multiplexing may be as effective as space-multiplexing for this kind of primed recognition task. In fact, it has been specifically suggested that image retrieval interfaces using Rapid Serial Visual Presentation (RSVP) may perform significantly better than parallel presentation even on a desktop computer [2]. In our experiments, we did not find this to be true. An important difference between previous experiments and our own, we discovered, is that image search engines rank retrievals, and correct answers are more likely to occur early in the list of results. Thus we found that scrolling (and low RSVP presentation rates) led to better recognition of answers that occur early, but worse for answers that occur far down the list.

This split confounded the global effects that we had hypothesized, yet in itself is an important consideration for future interface designs, which must adapt as search technology improves.

The high bandwidth required by live streaming video greatly limits the number of clients that can be served by a source. In this work, we discuss and evaluate an architecture, called SpreadIt, for streaming live media over a network of clients, using the resources of the clients themselves. Using SpreadIt, we can distribute bandwidth requirements over the network. The key challenge is to allow an application level multicast tree to be easily maintained over a network of transient peers, while ensuring that quality of service does not degrade. We propose a basic peering infrastructure layer for streaming applications, which uses a redirect primitive to meet the challenge successfully. Through empirical and simulation studies, we show that SpreadIt provides a good quality of service, which degrades gracefully with increasing number of clients. Perhaps more significantly, existing applications can be made to work with SpreadIt, without any change to their code base.

The high bandwidth required by live streaming video greatly limits the number of clients that can be served by a source. In this work, we discuss and evaluate an architecture, called SpreadIt, for streaming live media over a network of clients, using the resources of the clients themselves. Using SpreadIt, we can distribute bandwidth requirements over the network. The key challenge is to allow an application level multicast tree to be easily maintained over a network of transient peers, while ensuring that quality of service does not degrade. We propose a basic peering infrastructure layer for streaming applications, which uses a redirect primitive to meet the challenge successfully. Through empirical and simulation studies, we show that SpreadIt provides a good quality of service, which degrades gracefully with increasing number of clients. Perhaps more significantly, existing applications can be made to work with SpreadIt, without any change to their code base.

The high bandwidth required by live streaming video greatly limits the number of clients that can be served by a source using unicast. An efficient solution is IP-multicast, but it suffers from poor deployment. Application-level multicast is being increasingly recognized as a viable alternative. In this work, we discuss and evaluate a tree-based overlay network called PeerCast that uses clients to forward the stream to their peers. PeerCast is designed as a live-media streaming solution for peer-to-peer systems that are populated by hundreds of autonomous, short-lived nodes. Further, we argue for the need to take end-host behavior into account while evaluating an application-level multicast architecture. An end-host behavior model is proposed that allows us to capture a range of realistic peer behavior. Using this model, we develop robust, yet simple, tree-maintenance policies. Through empirical runs and extensive simulations, we show that PeerCast provides good QoS, which gracefully degrades with the number of clients. We have implemented a PeerCast prototype, which is available for download.

An important application of XML is the interchange of electronic data (EDI) between multiple data sources on the Web. As XML data proliferates on the Web, applications will need to integrate and aggregate data from multiple source and clean and transform data to facilitate exchange. Data extraction, conversion, transformation. and integration are all well-understood database problems, and their solutions rely on a query language. We present a query language for XML, called XML-QL, which we argue is suitable for performing the above tasks. XML-QL is a declarative. `relational complete' query language and is simple enough that it can be optimized. XML-QL can extract data from existing XML documents and construct new XML documents.

Format: HTML Document (34K + pictures) .

Audience: Medical forensics, computer scientists. References: 15. Links: 1. Relevance: Low-Medium. Abstract: Describes a system for collaboration among coroners. Focuses on issues of access rights-people in different roles (lead examiner, toxicologist, judge) see different views of the same data (some fields are read- prote

cted). Initial prototype was under the ABC system of UNC, but new prototypes will be web-based. Also hope to incorporate tele-conferencing capabilities.

This paper describes exploratory research concerning the automatic assignment of educational standards to lesson plans. An information retrieval based solution was proposed, and the results of several experiments are discussed. Results suggest the optimal solution would be a recommender tool where catalogers receive suggestions from the system but humans make the final decision.

Three types of video surrogates - visual (keyframes), verbal (keywords/phrases), and visual and verbal - were designed and studied in a qualitative investigation of user cognitive processes. The results favor the combined surrogates in which verbal information and images reinforce each other, lead to better comprehension, and may actually require less processing time, The results also highlight image features users found most helpful. These findings will inform the interface design and video representation for video retrieval and browsing.

Information retrieval over the Internet increasingly requires the filtering of thousands of information sources. As the number and variety of sources increases, new ways of automatically summarizing, discovering, and selecting sources relevant to a user's query are needed. Pharos is a highly scalable distributed architecture for locating heterogeneous information sources. Its design is hierarchical, thus allowing it to scale well as the number of information sources increases. We demonstrate the feasibility of the Pharos architecture using 2500 Usenet newsgroups as separate collections. Each newsgroup is summarized via automated Library of Congress classification. We show that using Pharos as an intermediate retrieval mechanism provides acceptable accuracy of source selection compared to selecting sources using complete classification information, while maintaining good scalability. This implies that hierarchical distributed metadata and automated classification are potentially useful paradigms to address scalability problems in large-scale distributed information retrieval applications.

This implementation paper introduces principles for the information architecture of an educational digital library, principles that address the distinction between designing digital libraries for education and designing digital libraries for information retrieval in general. Design is a key element of any successful product. Good designers and their designs put technology into the hands of the user, making the produc'ts focus comprehensible and tangible through design. As straightforward as this may appear, the design of learning technologies is often masked by the enabling technology. In fact, they often lack an explicitly stated instructional design methodology. While the technologies are important hurdles to overcome, we advocate learning systems that empower education-driven experiences rather than technology-driven experiences. This work describes a concept for a digital library for science, mathematics, engineering and technology education (SMETE), a library with an information architecture designed to meet learners' and educators' needs. Utilizing a constructivist model of learning, the authors present practical approaches to implementing the information architecture and it technology underpinnings. The authors propose the specifications for the information architecture and a visual design of a digital library for communicating learning to the audience. The design methodology indicates that a scenario-driven design technique sensitive to the contextual nature of learning offers a useful framework for tailoring technologies that help empower, not hinder, the educational sector.

This paper describes an evaluation model for a digital library tool, the Instructional Architect, which enables users to discover, select, reuse, sequence, and annotate digital library learning objects. By documenting our rapid-prototyping, iterative, and user-centered approach for evaluating a digital library service, we provide a model and set of methods that other developers may wish to employ. In addition, we provide preliminary results from our studies.

This paper presents a new technique for dealing with broken characters, one of the major challenges in the optical character recognition (OCR) of degraded historical printed documents. A technique based on graph combinatorics is used to rejoin the appropriate connected components. It has been applied to real data with successful results.

This paper presents a new toolkit for the creation of customized structured document recognition applications by domain experts. This open- source system, called Gamera, allows a user, with particular knowledge of the documents to be recognized, to combine image processing and recognition tools in an easy to use, interactive, graphical scripting environment. Gamera is one of the key technology components in a proposed international project for the digitization of diverse types of humanities documents.

As more information resources become accessible using computers, our digital interfaces to those resources need to be appropriate for all people. However when it comes to digital libraries, the interfaces have typically been designed for older children or adults. Therefore, we have begun to develop a digital library interface developmentally appropriate for young children (ages 5-10 years old). Our prototype system we now call SearchKinds offers a graphical interface for querying, browsing and reviewing search results. This paper describes our motifation for the research, the design partnership we established between children and adults, our design process, the technology outcomes of our current work, and the lessons we have learned.

This paper describes an ongoing collaborative effort across digital library and scientific communities in the UK to improve access to research data. A prototype demonstrator service supporting the discovery and retrieval of detailed results of crystallography experiments has been deployed within an Open Archives digital library service model. Early challenges include the understanding of requirements in this specialized area of chemistry and reaching consensus on the design of a metadata model and schema. Future plans encompass the exploration of commonality and overlap with other schemas and across disciplines, working with publishers to develop mutually beneficial service models, and investigation of the pedagogical benefits. The potential improved access to experimental data to enrich scholarly communication from the perspective of both research and learning provides the driving force to continue exploring these issues.

A main citation for LSI. It explains roughly how it works.

Unlike distributed transactions, mobile transactions do not originate and end at the same site. The implication of the movement of such transactions is that classical atomicity, concurrency and recovery solutions must be revisited to capture the movement behavior. As an effort in this direction, we define a model of mobile transactions by building on the concepts of split transactions and global transactions in a multidatabase environment. Our view of mobile transactions, called kangaroo transactions, incorporates the property that transactions in a mobile computing system hop from one base station to another as the mobile unit moves through cells. Our model is the first to capture this movement behavior as well as the data behavior which reflects the access to data located in databases throughout the static network. The mobile behavior is dynamic and is realized in our model via the use of split operations. The data access behavior is captured by using the idea of global and local transactions in a multidatabase system.

Computing metaphors have become an intricate part of information systems design. Yet, they are deeply rooted in cultural practices. This paper presents an investigation of the cross-cultural use and usability of the library metaphor in digital libraries. The study examines the relevant features of the Maori culture in New Zealand, their form of knowledge transfer and their use of real world and digital libraries. On this basis the paper points out why and when the library metaphor fails Maori and other indigenous users and how this knowledge can contribute to the improvement of future designs.

This paper describes the development of a communication aid for people with profound physical disabilities, people who cannot communicate verbally, and who cannot use conventional communication tools. The Greenstone digital library software has been used to construct a digital library of common conversational expressions. A case study approach was adopted, and the target user for this particular digital library was a local high school student. Tailoring the digital library's contents to this user entailed identifying physical accessibility considerations for her, developing a suitable mode of interaction with the digital library software, populating the digital library with appropriate expressions for her, and evaluating the digital library interface. Evaluation involved both a qualitative user evaluation session and a quantitative analysis of the time and effort required to use each of three proposed searching interfaces.

The field of music provides an interesting context for the development of digital library systems due to the variety of information formats used by music students and scholars. The VARIATIONS digital library project at Indiana University currently delivers online access to sound recordings from the collections of IU's William and Gayle Cook Music Library and is developing access to musical score images and other formats. This paper covers the motivations for the creation of VARIATIONS, an overview of its operation and implementation, user reactions to the system, and future plans for development.

With the increasing technical sophistication of both information consumers and providers, there is increasing demand for more meaningful experiences of digital information. We present a framework that separates digital object experience, or rendering, from digital object storage and manipulation, so the rendering can be tailored to particular communities of users. Our framework also accommodates extensible digital object behaviors and interoperability. The two key components of our approach are 1) exposing structural metadata associated with digital objects - metadata about labeled access points within a digital object and 2) information intermediaries called context brokers that match structural characteristics of digital objects with mechanisms that produce behaviors. These context brokers allow for localized rendering of digital information stored externally.

Resource discovery in a distributed digital library poses many challenges, one of which is how to choose search engines for query distribution. In this paper, we describe a federated, distributed digital library architecture and introduce the notion of a query mediator as a digital library service responsible for selecting among available search engines, routing queries to those search engines, and aggregating results. We examine operational data from the NCSTRL digital library, focusing on two characteristics of distributed resource discovery: availability (will a search engine respond within a time limit) and response time (how quickly will a search engine respond, given that it does respond) and distinguishing between the query mediator view of these characteristics and the indexer view. We also examine the accuracy of predictions we made of QM-view availability and response times of search engines.

In this paper, we describe an evaluation study of a self-archiving service for the Brazilian Digital Library of Computing (BDBComp). We conducted an extensive usability experiment with several potential users, including graduate students, professors, and archivists/librarians. The results of the study are described and analyzed, following sound statistical principles.

Format: HTML Document (15K).

Audience: Readers of Dr. Linn's article, lawyers. References: 5. Links: 0. Relevance: Low-medium. Abstract: Discusses the differences between a common carrier, distributor, and publisher. Also discusses trade secrets, fair use.

Format: HTML Document (6K) .

Audience: DL '94 officials & attendees. References: 0 . Links: 1 . Relevance: Low. Abstract: Queensland U's interest in attending DL '94. Some statistics on current and expected use of networked information servers.

This paper describes ongoing research into the application of machine learning techniques for improving access to governmental information in complex digital libraries. Under the auspices of the GovStat Project (http://www.ils.unc.edu/govstat), our goal is to identify a small number of semantically valid concepts that adequately spans the intellectual domain of a collection. The goal of this discovery is twofold. First we desire a principled aid to information architects. Second, automatically derived document-concept relationships are a necessary precondition for real-world deployment of many dynamic interfaces. The current study compares concept learning strategies based on three document representations: keywords, titles, and full-text. In statistical and user-based studies, human-created keywords provide significant improvements in concept learning over both title-only and full-text representations.

Ibiblio is a digital library whose materials are submitted and maintained by volunteer contributors. This study analyzes the emergence of hyperlinked structures within the ibiblio collection. In the context of ibiblio, we analyze the suitability of Barabasi's model of preferential attachment to describe the distribution of incoming links. We find that the degree of maintainer activity for a given site (as measured by the voluntary development of descriptive metadata) is a stronger link count predictor for ibiblio than is a site's age, as the standard model predicts. Thus we argue that the efforts of ibiblio's contributors positively affect the popularity of their materials.

Format: HTML Document (30K) .

Audience: Non-technical, social science, work flow. References: 16. Links: 1. Relevance: Low. Abstract: Case study of customer service organization that uses Lotus Notes. Discusses importance of face-to-face, informal communication, human information mediators.

Format: Compressed PostScript().

Format: HTML Document (30K + picture) .

Audience: General . References: 12. Links: 3. Relevance: Medium. Abstract: News wire stories on disasters are annotated with facts that relate to the reader's local region. (eg, casualties are cast as a multiple of the hometown population). Readers from different locales see different aug

mentations. Frames triggered by disaster keywords are filled in with relevant material, which is then personalized.

They go beyond retrieval, to include updates and other ops

A collection of papers that introduce blackboard systems, that provide a historical perspective of blackboard systems, that evaluate the contributions made by different systems, and that illustrate by example the range of blackboard applications and implementations.

Format: HTML Document (13K + pictures) .

Audience: Multimedia developers, computer scientists. References: 8. Links: 1. Relevance: Medium-Low. Abstract: Describes a rights management system for multimedia objects called LicensIt. Wrapper around object includes information about author, rights required, and digital signature to verify authenticity. Viewing objects i

s through special LicensIt viewers or through commercial applications with LicenseIt plug-ins.

The high publication rate of scholarly material makes searching and browsing an inconveninet way to keep oneself up-to-date. Instead of being the active part in information access, researchers want to be notified whenever a new paper in one's research area is published. While more and more publishing houses or portal sites offer notification services this approach has several disadvantages. We introduce the Hermes alerting service, a service that integrates a variety of different information providers making their heterogeneity transparent for the users. Hermes offers sophisticated filtering capabilities preventing the user from drowning in a flood of irrelevant information. From the user's point of view it integrates the providers into a single source. Its simple provider interface makes it easy for publishers to join the service and thus reaching the potential readers directly. This paper presents the architecture of the Hermes service and discusses the issues of heterogeneity of information sources. Furthermore, we discuss the benefits and disadvantages of message-oriented middleware for implementing such a service for digital libraries.

As large-scale collection of medical videos comes into view, there is an urgent need to develop semantic medical video classification techniques and enable video retrieval at the semantic level. However, most existing batch-based classifier training techniques still suffer from context and concept uncertainty problems when only a limited number of labeled training samples are available. To address the context and concept uncertainty problems, we have proposed a novel framework by integrating large-scale unlabeled samples with a limited number of labeled samples to enable more effective feature subset selection, parameter estimation and model selection. Specifically, this framework includes: (a) A novel multimodal context integration and semantic video concept interpretation framework; (b) A novel classifier training technique by integrating feature subset selection, parameter estimation and model selection seamlessly in a single algorithm to address the context uncertainty problem over time; (c) A cost-sensitive semantic video classification framework to address the concept uncertainty problem. Our experimental results in a certain medical education video domain have also been provided a convincing proof of our conclusions.

Format: Compressed PostScript().

This article describes the Compus visualization system that assists in the exploration and analysis of structured document copora encoded in XML. Compus has been developed for and applied to a corpus of 100 French manuscript letters of the 16th century, transcribed and encoded for scholarly analysis using the recommendations of the Text Encoding Initiative. By providing a synoptic visualization of a corpus and allowing for dynamic queries and structural transformations, Compus assists researchers in finding regularities or discrepancies, leading to a higher level analysis of historic source. compus can be used with other richly encoded text copora as well.

At the NASA Langley Research Center, the Earth Radiation Budget Experiment (ERBE) Data Management Team and the Atmodpheric Sciences Data Center are developing a digital collection for the ERBE project. The main goal is long-term preservation of a comprehensive information environment. The secondary goal is to provide a context for these data products by centralizing the 25 year research project's scattered information elements. The development approach incorporates elements of rapid prototyping and user-centered design in a standards-based implementation. A working prototype is in testing with a small number of users.

Format: Compressed PostScript().

Format: HTML Version (25K)> . Audience: Publishers. References: 0. Links: 0. Relevance: Low-Medium. Abstract: The president of the Online Bookstore gives her suggestions for other publishers to succeed in the digital age. Think creatively about things that were not possible in paper. Add value by licensing content then giv

ing people a framework through which to think about it. Quote from Gregory Rawlins, a computer science professor at Indiana University: If you're not part of the steamroller, you're part of the road.

Format: HTML Document (18K) . Audience: Publishers. References: 0. Links: 0. Relevance: Low. Abstract: Electronic publishing will need to account for the distributed nature of the Internet. Roles for a publisher include: Imprimatur of quality, content filter. Verifying authenticity of the files. Creating context ar

ound core content; Developing and maintaining an equitable royalty system based on number of accesses. Customizing the content for the readers.

Format: HTML Document (30K) . Audience: General public, publishers. References: 0. Links: 0. Relevance: low. Abstract: One person's view on the future of publishing and books. Time to press has decreased; non-linear thinking is encouraged; people use on-line resources differently than traditional books; piracy is not likely to be a

big problem; publishers still needed to publicize.

Format: HTML Document (15K) . Audience: Publishers, scholars, authors. References: 0. Links: 0. Relevance: low. Abstract: An MIT Press director gives her position on copyrights: Journal publishers are essential because they 1) take care of requests for reprints, etc and 2) provide the filter which full-text on-line services need to determine quality. Individual authors or their institutions could not do these economically. Does suggest some changes to current law, like allowing authors to copy for their own classes without fee.

Explores and uncovers a wide range of issues surrounding computer-augmented environments. The Chameleon prototype and a set of computer-augmented applications are described. Chameleon is a prototype system under development at the University of Toronto. It is part of an investigation on how palmtop computers designed with a high-fidelity monitor can become spatially aware of their location and orientation and serve as bridges or portholes between computer-synthesized information spaces and physical objects. In this prototype design, a 3D input controller and an output display are combined into one integrated unit.

We are exploring how virtual reality theories can be applied toward palmtop computers. In our prototype, called the Chameleon, a small 4-inch hand-held monitor acts as a palmtop computer with the capabilities of a Silicon graphics workstation. A 6D input device and a response button are attached to the small monitor to detect user gestures and input selections for issuing commands. An experiment was conducted to evaluate our design and to see how well depth could be perceived in the small screen compared to a large 21-inch screen, and the extent to which movement of the small display (in a palmtop virtual reality condition) could improve depth perception, Results show that with very little training, perception of depth in the palmtop virtual reality condition is about as good as corresponding depth perception in a large (but static) display. Variations to the initial design are also discussed, along with issues to be explored in future research, Our research suggests that palmtop virtual reality may support effective navigation and search and retrieval, in rich and portable information spaces.

Deals with prioritizing queries to information sources on the web, i.e., by first querying the ones more likely to be relevant.

Format: HTML Document (10K) . Audience: Corporate librarians. References: 0. Links: 1. Relevance: Low. Abstract: Discusses the role of a Knowledge Manager (formerly corporate librarian. In particular, finding and organizing new networked information resources.

LSI study to show how well it predicts interestingness of newsgroup articles.

With the increasing availability of information in electronic form, it becomes more important and feasible to have automatic methods to filter information. The results of an experiment aimed at determining the effectiveness of four information-filtering methods in the domain of technical reports are presented. The experiment was conducted over a six-month period with 34 users and over 150 new reports published each month. Overall, the authors conclude that filtering methods show promise for presenting personalized information.

Format: Compressed PostScript().

Format: HTML Document (52K).

Audience: Publishers, layperson, non-technical. References: 0. Links: 0. Relevance: Low. Abstract: Discussion of the CUPID project, basically just-in-time printing, (of textbooks, e.g.) at trusted printshops. No terminal display, hardcopy only.

As an increasing number of digital library projects embrace the harvesting of item-level descriptive metadata, issues of description granularity and concerns about potential loss of context when harvesting item-level metadata take on greater significance. Collection-level description can provide added context for item-level metadata records harvested from disparate and heterogeneous providers. This paper describes an ongoing experiment using collection-level description in concert with item-level metadata to improve quality of search and discovery across an aggregation of metadata describing resources held by a consortium of large academic research libraries. We present details of approaches implemented so far and preliminary analyses of the potential utility of these approaches. The paper concludes with a brief discussion of related issues and future work plans.

We describe our experience implementing orthogonal extensions to the existing WWW user interface, to support user control of intelligent services. Our extensions are orthogonal in that they provide an interface to a service, which complements the Web browsing experience but is independent of the content of any particular site. We base our experiments on the TranSend service at UC Berkeley, which performs lossy compression on inline images to accelerate dialup Web access for a community of 25,000 subscribers. The service keeps a separate preferences profile for each user, which allows each user to vary the aggressiveness of lossy compression, selectively turn off the service for certain pages, and select the type of interface provided for refinement of degraded (lossily compressed) content. We are exploring three technologies for implementing the TranSend service interface: HTML decoration, Java and JavaScript.

The Pythia proxy mechanism provides three important orthogonal benefits to World Wide Web (WWW) clients. (1) Real-time distillation and refinement, guided by statistical models, allow the user to bound latency and exercise explicit control over bandwidth that may be scarce and expensive (e.g. a metered cellular phone service). (2) Transcoding to a representation understood directly by the client system may improve rendering on the client or result in a representation that can be transmitted more efficiently. (3) Knowledge of client display constraints allows content to be optimized for rendering on the client. Users have commented that even the prototype version of Pythia provides a qualitative increase of about 5 times when surfing the World Wide Web over PPP (Point-to-Point Protocol) with a 14.4 kbit/s modem. These are the same users that previously turned image loading off completely in order to make surfing bearable. With the continued growth of the WWW, the benefits afforded by proxied services like Pythia will represent increasingly significant added value to end users and content providers alike. Pythia is the first fruit of a comprehensive research agenda aimed at implementing and deploying such services.

The explosive growth of the Internet and the proliferation of smart cellular phones and handheld wireless devices is widening an already large gap between Internet clients. Clients vary in their hardware resources, software sophistication, and quality of connectivity, yet server support for client variation ranges from relatively poor to none at all. In this paper we introduce some design principles that we believe are fundamental to providing meaningful Internet access for the entire range of clients. In particular, we show how to perform on-demand datatype-specific lossy compression on semantically typed data, tailoring content to the specific constraints of the client. We instantiate our design principles in a proxy architecture that further exploits typed data to enable application-level management of scarce network resources. Our proxy architecture generalizes previous work addressing all three aspects of client variation by applying well-understood techniques in a novel way, resulting in quantitatively better end-to-end performance, higher quality display output, and new capabilities for low-end clients.

Mobile computing is now at the stage where cell phones were 5-7 years ago. Laptops are frequently the choice of telecommuters who put in significant amounts of time both at home and at the office, but there is a growing group of mobile users who work from more than two locations and who expect to perform their full job responsibilities using a laptop that rarely returns to the main office. Although a mobile PC can be used without ever connecting to a network, they are typically connected with or without wires. Wired systems are most common and generally use modems with the dial-up lines found in homes or hotels. Wireless connections are increasing in popularity and use cellphone-like radio links to send and receive information, but mobile computing is not just about wireless connections; it is also about using your laptop in a hotel (in any country), at home, in a branch office or at a customer site. Using a laptop in those locations frequently reduces your interaction speed and range of functions to an unacceptable level, but recent improvements have attacked these problems. To really feel the freedom offered by mobile computing, imagine setting up overseas in your customer's spare office and working as if you were in your own office. Imagine being in a foreign country and not having to load printer drivers for each printer and load US fonts for each job in order to produce properly printed output. If this sort of future appeals to you, you're not alone. A likely and growing group is people who already use laptops. In 1996, laptops comprised 30-35 are forecast to be manufactured.

The Web provides access to a wide variety of information but much of this information is fluid; it changes, moves, and occasionally disappears. Bookmarks, paths over Web pages, and catalogs like Yahoo! are examples of page collections that can become out-of-date as changes are made to their components. Maintaining these collections requires that they be updated continuously. Tools to help in this maintenance require an understanding of what changes are important, such as when pages no longer exist, and what changes are not, such as when a visit counter changes. We performed a study to look at the effect of the type and quantity of change on people's perception of its importance. Subjects were presented pairs of Web pages with changes to either content (e.g., text), structure (e.g., links), or presentation (e.g., colors, layout). While changes in content were the most closely connected to subjects perceptions of the overall change to a page, subjects indicated a strong desire to be notified of structural changes. Subjects only considered the simultaneous change of many presentation characteristics as important.

Increasingly, digital libraries are being defined that collect pointers to World-Wide Web based resources rather than hold the resources themselves. Maintaining these collections is challenging due to distributed document owership and high fluidity. Typically a collection's maintainer has to assess the relevance of changes with little system aid. In this paper, we describe the Walden's Paths Path Manager, which assists a maintainer in discovering when relevant changes occur to linked resources. The approach and system design was informed by a study of how humans perceive changes of Web pages. The study indicated that structural changes are key in determining the overall change and that presentation changes are considered irrelevant.

Text categorization is typically formulated as a concept learning problem where each instance is a single isolated document. In this paper we are interested in a moe general formulation where documents are organized as page sequences, as naturally occurring in digital libraries of scanned books and magazines. We describe a method for classifying pages of sequential OCR text dosuments into one of several assigned categories and suggest that taking into account contextual information provided by the whole page sequence can significantly improve classification accuracy. The proposed architecture relies on hidden Markov models whose emissions are bag-of-words according to a multinomial word event model, as in the generative portion of the Naive Bayes classifier. Our results on a collection of scanned journals from the Making of America project contirm the importance of using whole page sequences. Empirical evaluation indicates that the error rate (as obtained by running a plain Naive Bayes classifier on isolated page) can be roughly reduced by hald if contextual information is incorporated.

Format: HTML Document (10K) . Audience: Legislators. References: 0. Links: 0. Relevance: low. Abstract: AAUP believes the current copyright law is sufficient for use in the networked world. The copyright provides a valuable service in designating the final (peer-reviewed) copy. Reprint fees, contracts, and copy prot

ection should not be mandated but handled by the copyright holders.

We have created three testbeds of web data for use in controlled experiments in collection modeling. This short paper examines the applicability of Ziff's and Heaps' laws as applied to web data. We find extremely close agreement between observed vocabulary growth and Heaps' law. We find reasonable agreement with Ziff's law for medium to low frequency terms. Ziff's law is a poor predictor for high frequency terms. These findings hold for all three testbeds although we restrict ourselves to one here due to space limitations.

Content-based image retrieval uses features that can be extracted from the images themselves. Using more than one representation of the images in a collection can improve the results presented to a user without changing the underlying feature extraction of search technologies. We present an example of this multiple viewpoint approach, multiple image channels. and discuss its advantages for an image-seeking user. This approach has also been shown to dramatically improve retrieval effectiveness in content-based image retrieval systems.

This is the WATERS paper.

Format: Not Yet Online.

Automatic text classification is an important operational problem in digital library practice. Most text classification efforts so far concentrated on developing centralized solutions. However, centralized classification approaches often are limited due to constraints on knowledge and computing resources. In addition, centralized approaches are more vulnerable to attacks or system failures and less robust in dealing with them. We present a de-centralized approach and system implementation (named MACCI) for text classification using a multi-agent framework. Experiments are conducted to compare our multi-agent approach with a centralized approach. The results show multi-agent classification can achieve promising classification results while maintaining its other advantages.

Archived biomedical literature and sequence data are growing rapidly. OAI-PMH provides a convenient way for data sharing, but it has not been tested in the biomedical domain, especially in dealing with different types of data, such as protein, and gene sequences. We built four individual OAI-PMH repositories based on different biomedical resources. Using the harvested data from the four repositories we created an integrated OAI-PMH repository, which hosts the linked literature and sequence data in a single place.

Information retrieval (IR) functions serve a critical role in many digital library systems. There are numerous mature IR algorithms that have been implemented and it will be a waste of resources and time to re-implement them. Those IR algorithms can be modulated and composed through the framework of web services. Web services in IR domain have not been widely tested. Concept extraction is an important area in traditional IR. We demonstrated that it can be easily adopted as IR web services and can be accessed in multiple ways. For the IR web services, we take advantage of a term representation database which was created as a result of a previous digital library project containing 31,928,892 terms found on 49,602,191 pages of the Web.

Multi-user distributed applications running on heterogeneous networks must be able to display user interface components on several platforms. In wide-area public networks, such as the Internet, the mix of platforms and participants in an application will occur dynamically; the user interface will need to coexist with environments completely uncontrolled by the designer. We have dealt with this issue by considering user interfaces as a kind of document specifying the application`s requirements and adopting SGML technology to process them locally. This approach provides new flexibility, with implications for the design of network browsers, such as those of the World Wide Web, and leads to an interesting class of active documents.

Format: HTML Document (4K) .

Audience: Librarians and Digital Library researchers. References: 0. Links: 1. Relevance: Low. Abstract: A view on the role in classification, reference, ensuring access, and collection development for the librarian of a digital library.

Format: HTML Document (31K) .

Audience: Authors, Developers, slightly technical. References: 43. Links: 1. Relevance: Low-Medium. Abstract: Discussion of SGMLs, their motivation, research issues, how they might be extended to non-text objects. Distinction between content & presentation

Format: Not yet online.

Format: HTML Document (40K + picture) . Audience: technical, computer scientists with some knowledge of computational linguistics. References: 31. Links: 1. Relevance: Medium-Low. Abstract: Discusses the challenges in indexing/searching large databases. Argues for a bootstrapping/machine learning approach to locate words in related contexts (surrounding words). Suggests specific data structures. Discusses tradeoffs between accuracy & speed, and scaling problems.

Existing peer-to-peer search networks generally fall into two categories: Gnutella-style systems that use arbitrary topology and rely on controlled flooding for search, and systems that explicitly build an underlying topology to efficiently support a distributed hash table (DHT). In this paper, we propose a hybrid scheme for building a peer-to-peer lookup service over arbitrary network topology. Specifically, for each node in the search network, we build a small DHT consisting of nearby nodes and then provide an intelligent search mechanism that can traverse all the small DHTs. Our hybrid approach can reduce the nodes contacted for a lookup by an order of magnitude compared to Gnutella, allows rapid searching of nearby nodes through quick fan-out, does not reorganize the underlying overlay, and isolates the effect of topology changes to small areas for better scalability and stability.

Format: Not Yet Online.

Format: Compressed PostScript().

Format: HTML Document().

Format: HTML Document (29K) . Audience: slightly technical, generalist comfortable with technology. References: 23. Links: 1. Relevance: Medium-Low (but not mainstream DL). Abstract: Describes architecture of a system to retrieve & deliver video on demand. Indexing done by audio track or transcript. 100 hours of video to 20-30 users. Different compression modes depending on bandwidth of user co

nnection.

Format: Not yet online.

BoW is an on-line bibliographical repository based on a hierarchical concept index to which entries are linked. Searching in the repository should therefore return matching topics from the hierarchy, rather than just a list of entries. Likewise, when new entries are inserted, a search for relevant topics to which they should be linked is required. We develop a vector-based algorithm that creates keyword vectors for the set of competing topics at each node in the hierarchy, and show how its performance improves when domain-specific features are added (such as special handling of topic titles and author names). The results of a 7-fold cross validation on a corpus of some 3,500 entries with a 5-level index are hit ratios in the range of 89-95the misclassifications are indeed ambiguous to begin with.

Digital libraries have the potential to not only duplicate many of the services provided by traditional libraries but to extend them. Basic finding aids such as search and browse are common in most of today's digital libraries. But just as a traditional library provides more than a card catalog and browseable shelves of books, an effective digital library should offer a wider range of services. Using the traditional library concept of special collections as a model, in this paper we propose that explicitly defining sub-collections in the digital library-virtual collections-can benefit both the library's users and contributors and increase its viability. We first introduce the concept of a virtual collection, outline the costs and benefits for defining such collections, and describe an implementation of collection- level metadata to create virtual collections for two different digital libraries. We conclude by discussing the implications of virtual collections for enhancing interoperability and sharing across digital libraries, such as those that are part of the National SMETE Digital Library.

Discusses important issues related to agent-based software engineering, which was developed to create interaperable softwares.

Infomaster is an information integration system that provides integrated access to multiple distributed heterogeneous information sources on the Internet, thus giving the illusion of a centralized, homogeneous information system. We say that Infomaster creates a virtual data warehouse. The core of Infomaster is a facilitator that dynamically determines an efficient way to answer the user's query using as few sources as necessary and harmonizes the heterogeneities among these sources. Infomaster handles both structural and content translation to resolve differences between multiple data sources and the multiple applications for the collected data. Infomaster connects to a variety of databases using wrappers, such as for Z39.50, SQL databases through ODBC, EDI transactions, and other World Wide Web (WWW) sources. There are several WWW user interfaces to Infomaster, including forms based and textual. Infomaster also includes a programmatic interface and it can download results in structured form onto a client computer. Infomaster has been in production use for integrating rental housing advertisements from several newspapers (since fall 1995), and for meeting room scheduling (since winter 1996). Infomaster is also being used to integrate heterogeneous electronic product catalogs.

Format: HTML Document (40K).

Audience: Standards committees, general technologists, technical sections. References: 14 footnotes. Links: 0 . Relevance: low-medium. Abstract: Discusses a header-based approach to identifying data streams, focusing on video domains. Gives a brief history of copyrights. Gives desiderata for standards/design to ensure interoperability, flexibility, extensib

ility, etc. Gives concrete examples of encoding used for certain applications.

Format: Not Yet Online.

Format: HTML Document (11K) . Audience: Image catalogers & users, politicians. References: 0. Links: 0. Relevance: low. Abstract: Argues that all of the search & retrieval issues for bibliographic records are worse for images. There are more of them, there's no standard for representation or indexing. Calls for creation of universal image database, a single standard for representation, and a standard license agreement for image owners.

Format: Not yet online.

The automatic document metadata extraction process is an important task in a world where thousands of documents are just one click away. Thus, powerful indices are necessary to support effective retrieval. The upcoming XML standard represents an important step in this direction as its semistructured representation conveys document metadata together with the text of the document. For example, retrieval of scientific papers by authors or affiliations would be a straightforward task if papers were stored in XML. Unfortunately, today, the largest majority of documents on the web are available in forms that do not carry additional semantics. Converting existing documents to a semistructured representation is time consuming and no automatic process can be easily applied. In this paper we discuss a system, based on a novel spatial/visual knowledge principle, for extracting metadata from scientific papers stored as PostScript files. Our system embeds the general knowledge about the graphic layout of a scientific paper to guide the metadata extraction process. Our system can effectively assist the automatic index creation for digital libraries.

Format: HTML Document (35K) .

Audience: DL researchers, workgroup attendees. References: 22. Links: 1. Relevance: Low-Medium. Abstract: The report of a working group on digital libraries. Defines terms, discusses a possible architecture in terms of resource managers and application enablers.

Digital library research is mostly focused on the generation of large collections of multimedia resources and state-of-the-art tools for their indexing and retrieval. However, digital libraries should provide more than advanced collection maintenance and retrieval services since the ultimate goal of any academic library is to serve the scholarly needs of its users. This paper begins by presenting a case for digital scholarship in which patrons perform all scholarly work electronically. A proposal is then made for patron-augmented digital libraries (PADLs), a class of digital libraries that supports the digital scholarship of its patrons. Finally, a prototype PADL (called Synchrony) providing access to video segments and associated textual transcripts is described. Synchrony allows patrons to search the library for artifacts, create annotations/original compositions, integrate these artifacts to form synchronized mixed text and video presentations and, after suitable review, publish these presentations into the digital library if desired. A study to evaluate the PADL concept and usability of Synchrony is also discussed. The study revealed that participants were able to use Synchrony for the authoring and publishing of presentations and that attitudes toward PADLs were generally positive.

An individual scientist, a collaborative team and a research network have a variety of document management needs in common. The levels of research organization, when viewed as nested tiers, represent boundaries across which information can flow openly if technology and metadata standards are partnered to provide an accessible, interoperable digital framework. The CERN Document System (CDS), implemented by a research partnership at the San Diego Supercomputer Center (SDSC), establishes a prototype tiered repository system. An ongoing exploration of existing scientific research information infrastructure suggests modifications to enable cross-tier and cross-domain information flow across what could be represented as a metadata grid.

use user annotations to help with filtering.

Tapestry is an experimental mail system developed at the Xerox Palo Alto Research Center. The system manages an in-coming stream of electronic documents, including E-mail, newswire stories and NetNews articles. The system implements a novel mechanism for collaborative filtering in which users annotate documents before the documents art filtered. Because annotations are not available at the time a new document arrives, the system supports continuous queries that examine the entire database of documents and take into account newly introduced annotations during the filtering process.

This paper describes an extension to the current Java security model called the Gateway and why it was necessary to create it. This model allows secure applications, such as those used in electronic commerce, to safely exchange data and interoperate without compromising each individual application's security. The Gateway uses digital signatures to enable application programming interfaces to authenticate their caller. JavaSoft is using the Gateway to create a new integrated open platform for financial applications called Java Electronic Commerce Framework. The JECF will be the foundation for electronic wallets, point of sale terminals, electronic merchant servers and other financial software. The Gateway model can also be used for access control in many multiple application environments that require trusted interaction between applications from multiple vendors. These applications include browsers, servers, operating systems, medical systems and smartcards.

Digital libraries (DLs) are among the most complex kinds of information systems, due in part to their intrinsic multi-disciplinary nature. Nowadays DLs are built within monolithic, tightly integrated, and generally inflexible systems - or by assembling disparate components together in an ad-hoc way, with resulting problems in interoperability and adaptability. More importantly, conceptual modeling, requirements analysis, and software engineering approaches are rarely supported, making it extremely difficult to tailor DL content and behavior to the interests, needs, and preferences of particular communities. In this paper, we address these problems. In particular, we present 5SL, a declarative language for specifying and generating domain- specific digital libraries. 5SL is based on the 5S formal theory for digital libraries and enables high-level specification of DLs in five complementary dimensions, including: the kinds of multimedia information the DL supports (Stream Model); how that information is structured and organized (Structural Model); different logical and presentational properties and operations of DL components (Spatial Model); the behavior of the DL (Scenario Model); and the different societies of actors and managers of services that act together to carry out the DL behavior (Societal Model). The practical feasibility of the approach is demonstrated by the presentation of a 5SL digital library generator for the MARIAN digital library system.

Structured or fielded metadata is the basis for many digital library services, including searching and browsing. Yet, little is known about the impact of using structure in the effectiveness of such services. In this paper, we investigate a key research question: do structured queries improve effectiveness in DL searching? To answer this question, we empirically compared the use of unstructured queries to the use of structured queries. We then tested the capability of a simple Bayesian network system, built on top of a DL retrieval engine, to infer the best structured queries from the keywords entered by the user. Experiments performed with 20 users working with a DL containing a large collection of computer science literature clearly indicate that structured queries, either manually constructed or automatically generated, perform better than their unstructured counterparts, in the majority of cases. Also, automatic structuring of queries appears to be an effective and viable alternative to manual structuring that may significantly reduce the burden on users.

We describe current efforts and developments building on our proposal for an XML log standard format for digital library (DL) logging analysis and companion tools. Focus is given to the evolution of formats and tools based on analysis of deployment in several DL systems and testbeds. Recent development of analysis tools also is discussed.

Format: Not Yet On-line.

Audience: Multimedia standards setters, developers - technical. References: 11. Relevance: Low. Abstract: Describes the MHEG standard being developed for multimedia objects and applications. Designed to deliver real-time interchange of multimedia objects over wide area networks.

We developed two photo browsers for collections with thousands of time-stamped digital images. Modern digital cameras record photo shoot times, and semantically related photos tend to occur in bursts. Our browsers exploit the timing information to structure the collections and to automatically generate meaningful summaries. The browsers differ in how users navigate and view the structured collections. We conducted user studies to compare the two browsers and a commercial image browser. Our results show that exploiting the time dimension and appropriately summarizing collections can lead to significant improvements. For example, for one task category, one of our browsers enabled a 33commercial browser. Similarly, users were able to complete 29when using this same browser.

Format: HTML Document (43K).

Audience: Non-technical, librarians. References: 13 notes. Links: 0. Relevance: Low. Abstract: Discussion of ensuring authenticity of documents, essentially just notarization.

Format: HTML Document(36K) .

Audience: Librarians. References: 23. Links: 8. Relevance: Low. Abstract: Discusses technical and organizational challenges which must be met to have a real digital research library. SIgnificant one is obtaining the institutional commitments to ensure longevity of the collection and acces

s to it. Discusses some of the required tasks (like cataloging, backup , authentication) at a high level.

As a foundation for designing computer-supported photograph management tools, we have been conducting focused experiments. Here, we describe our analysis of how people initially organize batches of familiar images. We asked 26 subjects in pairs to organize 50 images on a common horizontal table. Each pair then organized a different 50-image set on a computer table of identical surface area. The bottom-projected computer tabletop displayed our interface to several online, pile-based affordances we wished to evaluate. Subjects used pens to interact with the system. We highlight aspects of the computer environment that were notably important to subjects, and others that they cared about less than we had hypothesized. For example, a strong majority preferred computer-generated representations of piles to be grid- shaped over several alternatives, some of which mimicked the physical world closely, and others that used transparency to save space.

The popularity of on-line document databases has led to a new problem: finding which text databases (out or many candidate choices) are the most relevant to a user. Identifying the relevant databases for a given query is the text database discovery problem. The first part of this paper presents a practical solution based on estimating the result size of a query and a database. The method is termed GLOSS-Glossary of Servers Server. The second part of this paper evaluates the effectiveness of GLOSS based on a trace of real user queries. In addition, we analyze the storage cost of our approach.

Update anywhere-anytime-anyway transactional replication has unstable behavior as the workload scales up: a ten-fold increase in nodes and traffic gives a thousand fold increase in deadlocks or reconciliations. Master copy replication (primary copy) schemes reduce this problem. A simple analytic model demonstrates these results. A new two-tier replication algorithm is proposed that allows mobile (disconnected) applications to propose tentative update transactions that are later applied to a master copy. Commutative update transactions avoid the instability of other replication schemes

This is a comprehensive book on transaction processing. Chapter 3 introduces the concept of Fault Tolerance. Chapter 4 presents different transaction models. Chapters 5,6 give an overview of the functionality of the TP monitor. Chapters 7,8 describe concurrency control and its implementation. Chapter 9 give an overview of recovery and how to implement logs. Chapter 10,11 defines a transaction manager and how to implement it. Chapter 12 is a compedium of advanced transaction manager topics including heterogeneous commit coordinators, non-blocking commit coordinators, transfer of commit, optimization of 2-phase commit, and disaster recovery. The rest of the book describes in detail the low level implementation of a transaction processing system, and provides a survery of TP systems in the market.

Format: Not Yet On-line.

Audience: Vision/graphics researchers. References: 15. Links: . Relevance: Low. Abstract: Technical description of implementation of two techniques to retrieve images based on color histograms and edge maps. Implemented and tested on a small (48 image) database. Results mixed at best. Weaknesses identif

ied for future work.

In this paper we describe how we combined SDLIP and STARTS, two complementary protocols for searching over distributed document collections. The resulting protocol, which we call SDARTS, is simple yet expressible enough to enable building sophisticated metasearch engines. SDARTS can be viewed as an instantiation of SDLIP with metasearch-specific elements from STARTS. We also report on our experience building three SDARTS-compliant wrappers: for locally available plain-text document collections, for locally available XML document collections, and for external web-accessible collections. These wrappers were developed to be easily customizable for new collections. Our work was developed as part of Columbia University's Digital Libraries Initiative-Phase 2 (DLI2) project, which involves the departments of Computer Science, Medical Informatics, and Electrical Engineering, the Columbia University libraries, and a large number of industrial partners. The main goal of the project is to provide personalized access to a distributed patient-care digital library.

This paper exposes the concurrency control problem in groupware when it is implemented as a distributed system. Traditional concurrency control methods cannot be applied directly to groupware because system interactions include people as well as computers. Methods, such as locking, serialization, and their degree of optimism, are shown to have quite different impacts on the interface and how operations are displayed and perceived by group members. The paper considers both human and technical considerations that designers should ponder before choosing a particular concurrency control method. It also reviews the authors` work-in-progress designing and implementing a library of concurrency schemes in GROUPKIT, a groupware toolkit.

Format: HTML Document (7K)

Audience: Women writers, readers. References: 6. (though not in on-line version) Links: 1. Abstract: Argues that the non-linear nature of multimedia fits better with the more cyclical nature of female life, and the non-linear way that many women authors write. The new medium offers women a chance to fight the patria

rchy of tradition.

Format: HTML Document (22K) . Audience: Browser developers, content publishers. References: 7. Links: 13. Relevance: low. Abstract: Quote from conclusion HTML is inadequate. It lacks sufficient structural and formatting tags to even render certain kinds of fiction comprehensible much less aesthetic. HTML needs style sheets or improved formatting capabilities so that document designers can spare 20 million Internet users from adjusting everything themselves. The META tag in HTML level 2 can be exploited to implement a document typing system. We need to develop a hierarchy of do cument types to facilitate implementation of programs that automatically process Web documents. This type system must support multiple inheritance.

Format: HTML Document (34K) . Audience: slightly technical, funders, general technology. References: 14. Links: 1. Relevance: Low-Medium. Abstract: Describes U. of Tennessee's Digital Library proposal. Focuses on: representation, navigation, retrieval, display of information; performance & scalability; different user interfaces for different user groups. Semantic net concept representation.

Format: HTML Document(29K).

Audience: Computer scientists, specific & somewhat technical. References: 10. Links: 0. Relevance: Medium. Abstract: Secure copyrighted documents by transmitting them in an envelope which is the only way to view, print, etc. Periodic & per-use reverification with a server, possible chargeback info. PATENTS APPLIED FOR.

This paper discusses how to augment the World Wide Web with an open hypermedia service (Webvise) that provides structures such as contexts, links, annotations, and guided tours stored in hypermedia databases external to the Web pages. This includes the ability for users collaboratively to create links front parts of HTML Web pages they do not own and support for creating links to parts of Web pages without writing HTML target tags. The method for locating parts of Web pages can locate parts of pages across frame hierarchies and it also supports certain repairs of links that break due to modified Web pages. Support for providing links to/from parts of non-HTML data, such as sound and movie, will be possible via interfaces to plug-ins and Java-based media players. The hypermedia structures are stored in a hypermedia database, developed from the Devise Hypermedia framework, and the service is available on the Web via an ordinary URL. The best user interface for creating and manipulating the structures is currently provided for the Microsoft Internet Explorer 4.x browser through COM integration that utilizes the Explorer's DOM representation of Web-pages. But the structures can also be manipulated and used via special Java applets and a pure proxy server solution is provided for users who only need to browse the structures. A user can create and use the external structures as `transparency' layers on top of arbitrary Web pages, the user can switch between viewing pageswith one or more layers (contexts) of structures or without any external structures imposed on them.

HTML Document (24K) . Audience: Computer scientists, funders . References: 7. Links: 1. Relevance: low. Abstract: Describes a prototype system built using tools to access data from the federal budget. Argues that statistical, numerical data is fundamentally different from text and multimedia.

The Web mostly contains semi-structured information. It is, however, not easy to search and extract structural data hidden in a Web page. Current practices address this problem by (1) syntax analysis (i.e. HTML tags); or (2) wrappers or user-defined declarative languages. The former is only suitable for highly structured Web sites and the latter is time-consuming and offers low scalability. Wrappers could handle tens, but certainly not thousands, of information sources. In this paper, we present a novel information mining algorithm, namely KPS, over semi-structured information on the Web. KPS employs keywords, patterns and/or samples to mine the desired information. Experimental results show that KPS is more efficient than existing Web extracting methods.

Format: HTML Document(46K + pictures) .

Audience: Mathematicians, economists, statisticians, computer scientists. References: 37. Links: 13. Relevance: Low. Abstract: Proposes a web-based repository for software implementing time series analysis methods. Such a system would facilitate collaboration, verification of results, and would help build an experience base of which models worked well under which circumstances. Briefly describes the architecture, which requires method implementors to specify the methods in terms of Ypsilon (an abstract class system) classes.

We describe our effort to develop an ontology service on top of an educational digital library. The ontology is developed by relating library holdings to the educational concepts they refer to. The ontology system supports basic services like ontology-based search and complex services such as comparison of multiple curricula.

A novel algorithm for speech watermarking through parametric modeling is enhanced by inclusion of a quantified fidelity criterion. Watermarking is effected through solution of a set-membership filtering (SMF) problem, subject to an l(infinity) fidelity criterion in the signal space. The SMF approach provides flexibility in obtaining watermark solutions that trade-off watermark robustness and stegosignal fidelity.

This paper, describes our experience with the creation, indexing and providing access to a very large archive of videotaped oral histories- 116,000 hours of digitized interviews in 32 languages from 52,000 survivors, liberators, rescuers and witnesses of the Nazi Holocaust-and identifies a set of critical research issues in user requirement studies, automatic speech recognition, automatic classification, segmentation, and summarization, retrieval, and user interfaces that must be addressed if we are to provide full and detailed access to collections of this size.

Format: Compressed PostScript().

We describe the Computing Research Repository (CoRR), a new electronic archive for rapid dissemination and archiving of computer science research results. CoRR was initiated in September 1998 through the cooperation of ACM, LANL (Los Alamos National Laboratory) e-Print archive, and NCSTRL (Networked Computer Science Technical Reference Library. Through its implementation of the Dienst protocol, CoRR combines the open and extensible architecture of NCSTRL with the reliable access and well-established management practices of the LANL XXX e-Print repository. This architecture will allow integration with other e-Print archives and provides a foundation for a future broad-based scholarly digital library. We describe the decisions that were made in creating CoRR, the architecture of the CoRR/NCSTRL interoperation, and issues that have arisen during the operation of CoRR.

Format: Compressed PostScript().

Automatic metadata generation provides scalability and usability for digital libraries and their collections. Machine learning methods offer robust and adaptable automatic metadata extraction. We describe a support vector machine classification-based method for metadata extraction from the header part of the research papers and show that it outperforms other machine learning methods on the same task. The method first classifies each line of the header into one or more of the 15 classes. An iterative convergence procedure is then used to improve the line classification by using the predicted class labels of its neighbor lines in the previous round. Further metadata extraction is done by seeking the best chunk boundaries of each line. We found that discovery and use of the structural patterns of the data and domain based feature selection can improve the metadata extraction performance. An appropriate feature normalization also greatly improves the classification performance.

Due to name abbreviations, identical names, or name misspellings in publications or bibliographies (citations), an author may have multiple names and multiple authors may share the same name. Such name ambiguity affects the performance of document retrieval, web search, database integration, and may cause improper attribution to authors. This paper investigates two supervised learning approaches to disambiguate authors in the citations. One approach uses the naive Bayes probability model, a generative model; the other uses Support Vector Machines(SVMs) [?] and vector space representation of citations, a discriminative model. Both approaches utilize three types of citation attributes: co-author names, the title of the paper ,and the title of the journal or proceeding. We illustrate these two approaches on two types of data, one collected from the web, mainly publication lists from homepages, the other collected from the DBLP citation databases.

An author may have multiple names and multiple authors may share the same name simply due to name abbreviations, identical names, or name misspellings in publications or bibliographies (citations). This can produce name ambiguity which can affect the performance of document retrieval, web search, and database integration, and may cause improper attribution of credit. Proposed here is an unsupervised learning approach using K-way spectral clustering that disambiguates authors in citations. The approach utilizes three types of citation attributes: co-author names, paper titles, and publication venue titles. The approach is illustrated with 16 name datasets with citations collected from the DBLP database bibliography and author home pages and shows that name disambiguation can be achieved using these citation attributes.

We have developed a wearable, personalized digital library system, which unobtrusively records the wearer's part of a conversation, recognizes the face of the current dialog partner and remembers his/her voice. The next time the system sees the same person's face and hears the same voice, it can replay the audio from the last conversation in compressed form summarizing the names and major issues mentioned. Experiments with a prototype system show that a combination of face recognition and speaker identification can be effective for retrieving conversations.

Today's web is so huge and diverse that it arguably reflects the real world. For this reason, searching the web is a promising approach to find things in the real world. This paper presents NEXAS, an extension to web search engines that attempts to find real-world entities relevant to a topic. Its basic idea is to extract proper names from the web pages retrieved for the topic. A main advantage of this approach is that users can query any topic and learn about relevant real-world entities without dedicated databases for the topic. In particular, we focus on an application for finding authoritative people from the web. This application is practically important because once personal names are obtained, they can lead users from the web to managed information stored in digital libraries. To explore effective ways of finding people, we first examine the distribution of Japanese personal names by analyzing about 50 million Japanese web pages. We observe that personal names appear frequently on the web, but the distribution is highly influenced by automatically generated texts. To remedy the bias and find widely acknowledged people accurately, we utilize the number of servers containing a name instead of the number of web pages. We show its effectiveness by an experiment covering a wide range of topics. Finally, we demonstrate two examples and discuss possible applications.

We developed two browsers to support large personal photo collections on PDAs. Our first browser is based on a traditional, folder-based layout that utilizes either the user's manually created organization structure, or a system-generated structure. Our second browser uses a novel interface that is based on a vertical, zoomable timeline. This timeline browser does not require users to organize their photos, but instead, relies solely on system-generated structure. Our system creates a hierarchical structure of the user's photos by applying time-based clustering to identify subsets of photos that are likely to be related. In a user experiment, we compared users' searching and browsing performance across these browsers, using each user's own photo collection. Photo collection sizes varied between 500 and 3000 photographs. Our results show that our timeline browser is at least as effective for searching and browsing tasks as a traditional browser that requires users to manually organize their photos.

This paper describes a novel tool, SmartSkim, for content-based browsing or skimming of documents. The tool integrates concepts from passage retrieval and from interfaces, such as TileBars, which provide a compact overview of query term hits within a document. We base our tool on the concept of relevance profiling, in which a plot of retrieval status values at each word position of a document is generated. A major contribution of this paper is applying language modelling to the task of relevance profiling. We describe in detail the design of the SmartSkim tool, and provide a critique of the design. Possible applications of the tool are described, and we consider how an operational version of SmartSkim might be architected.

Though the Open Archive Initiative Protocol for Metadata Harvesting (OAI-PMH) is becoming the de facto standard for digital libraries, some of its predecessors are still in use. Although a limited number of Dienst repositories continue to be populated, others are precariously unsupported. The Dienst Open Archive Gateway (DOG) is a gateway between the OAI-PMH and the Dienst (version 4.1) protocol. DOG allows OAI-PMH harvesters to extract metadata records (in RFC-1807 or Dublin Core) from Dienst servers.

We examine multi-modal information retrieval from broadcast video where text can be read on the screen through OCR and speech recognition can be performed on the audio track. OCR and speech recognition are compared on the 2001 TREC Video Retrieval evaluation corpus. Results show that OCR is more important that speech recognition for video retrieval. OCR retrieval can further improve through dictionary-based post-processing. We demonstrate how to utilize imperfect multi-modal metadata results to benefit multi-modal information retrieval.

We propose a new analysis for searching images in video libraries that goes beyond simple image search, which compares one still image frame to another. The key idea is to expand the definition of an image to account for the variability in the sequence of video frames that comprise a shot. A first implementation of this method for a QBIC-like image search engine shows a clear improvement over still image search. A combination of the traditional still image search and the new video image search provided the overall best results on the TREC video retrieval evaluation data.

Format: HTML Document().

In the original PageRank algorithm for improving the ranking of search-query results, a single PageRank vector is computed, using the link structure of the Web, to capture the relative importance of Web pages, independent of any particular search query. To yield more accurate search results, we propose computing a set of PageRank vectors, biased using a set of representative topics, to capture more accurately the notion of importance with respect to a particular topic. By using these (precomputed) biased PageRank vectors to generate query-specific importance scores for pages at query time, we show that we can generate more accurate rankings than with a single, generic PageRank vector. For ordinary keyword search queries, we compute the topic-sensitive PageRank scores for pages satisfying the query using the topic of the query keywords. For searches done in context (e.g., when the search query is performed by highlighting words in a Web page), we compute the topic-sensitive PageRank scores using the topic of the context in which the query appeared.

A frozen 18.5 million page snapshot of part of the Web has been created to enable and encourage meaningful and reproducible evaluation of Web search systems and techniques. This collection is being used in an evaluation framework within the Text Retrieval Conference (TREC) and will hopefully provide convincing answers to questions such as, Can link information result in better rankings?`, Do longer queries result in better answers?`, and, Do TREC systems work well on Web data? The snapshot and associated evaluation methods are described and an invitation is extended to participate. Preliminary results are presented for an effective comparison of six TREC systems working on the snapshot collection against five well-known Web search systems working over the current Web. These suggest that the standard of document rankings produced by public Web search engines is by no means state-of-the-art.

The field of information retrieval has traditionally focused on textbases consisting of titles and abstracts. As a consequence, many underlying assumptions must be altered for retrieval from full-length text collections. This paper argues for making use of text structure when retrieving from full text documents, and presents a visualization paradigm, called TileBars, that demonstrates the usefulness of explicit term distribution information in Boolean-type queries. TileBars simultaneously and compactly indicate relative document length, query term frequency, and query term distribution. The patterns in a column of TileBars can be quickly scanned and deciphered, aiding users in making judgments about the potential relevance of the retrieved documents.

Discusses the issues related to semantic interoperability. The purposes are to indicate why semantic interoperability is so hard to achieve, and to suggest that repository technology can provide the beginnings of help to make it easier.

Format: HTML Document (34K + pictures) .

Audience: Instructors, students, textbook authors and publishers. References: 15. Links: 6. Relevance: Low-medium. Abstract: A very broad but shallow description of the textbook production business. Argues for a distributed author model, but with publishers that still piece together textbooks from the contributions of instructors, students

, etc. CD-ROM versions in addition to on-line. Authors paid according to their contribution. Briefly mentions administration, intellectual property issues. Longer example of physics/engineering systems demo.

Format: HTML Document (71K).

Audience: Public policy, lawyers, and developers. References: 47 notes. Links: 0. Relevance: Medium. Abstract: Focusing primarily on intellectual property, this article covers a lot of ground. Briefly describes the CNRI copyright management project, argues for permission headers that describe how each of the various protected rights (viewing, copying, preparing derivative works, etc) can be in the header, along with economic information. Describes whether digitally signed contracts are likely to be legally enforceable (they probably are), and under what ci

rcumstances electronic records are court-admissable (when they are generated as a regular course of business, and there's no reason to doubt them). Argues against general encryption, too expensive & inconsistent with the open market o

f ideas. Seeks legal protection commensurate with the value of a transaction.

Recent research has studied how to measure the size of a search engine, in terms of the number of pages indexed. In this paper, we consider a different measure for search engines, namely the quality of the pages in a search engine index. We provide a simple, effective algorithm for approximating the quality of an index by performing a random walk on the Web, and we use this methodology to compare the index quality of several major search engines.

Introduces the NC spec and discusses its chances

This paper introduces the shark search algorithm, a refined version of the first dynamic Web search lgorithms, the fish search. The shark-search has been embodied into a dynamic Web site mapping that enables users to tailor Web maps to their interests. Preliminary experiments show significant improvement over the original fish-search algorithm.

Recording on digital objects (e. g. reports, forms, contracts, mail-order catalogs, source code, manual pages, email, spreadsheets, menus) the interaction events that comprise their use makes it possible on future occasions, when the objects are used again, to display graphical abstractions of the accrued histories as parts of the objects themselves. For example, co-authors of a report can see stable and unstable sections (lines of text are marked by recency of changes or amount of editing) and identify who has written what and when. In the case of reading documentation, a reader can see who else has previously read a particular section of interest. While using a spreadsheet to refine a budget, the count of edits per spreadsheet cell can be mapped onto grayscale to give an impression of which budget numbers have been reworked the most and least. Or in the context of learning unfamiliar menu selections in a new piece of software, the menu itself can depict the distribution statistics of colleagues' previous menu selections in the same or similar contexts. There are many existing computational devices that hint at the prospect of history-enriched digital objects. Automatic change-bars, citation indices, and download counts on computer bulletin boards are examples. In fact, for the last thirteen years, members of our lab have been able to request AP News articles by specifying a minimum number of previous readers and thus easily retrieve articles that colleagues have chosen to read.

When making a choice in the absence of decisive first-hand knowledge, choosing as other like-minded, similarly-situated people have successfully chosen in the past is a good strategy-in effect, using other people as filters and guides: filters to strain out potentially bad choices and guides to point out potentially good choices. Current human-computer interfaces largely ignore the power of the social strategy. For most choices within an interface, new users are left to fend for themselves and if necessary, to pursue help outside of the interface. We present a general history-of-use method that automates a social method for informing choice and report on how it fares in the context of a fielded test case: the selection of videos from a large set. The positive results show that communal history-of-use data can serve as a powerful resource for use in interfaces.

In this paper, we study the problem of constructing and maintaining a large shared repository of web pages. We discuss the unique characteristics of such a repository, propose an architecture, and identify its functional modules. We focus on the storage manager module, and illustrate how traditional techniques for storage and indexing can be tailored to meet the requirements of a web repository. To evaluate design alternatives, we also present experimental results from a prototype repository called WebBase, that is currently being developed at Stanford University. Keywords : Repository, WebBase, Architecture, Storage management

The rapid growth of scholarly information resources available in electronic form and their organisation by digital libraries is proving fertile ground for the development of sophisticated new services, of which citation linking with be one indispensable example. Many new projects, partnerships and commercial agreements have been announced to build citation linking applications. This paper describes the Open Citation (OpCit) project, which will focus on linking papers held in freely accessible eprint archives such as the Los Alamos physics archives and other distributed archives, and which will build on the work of the Open Archives initiative to make the data held in such archives available to compliant services. The paper emphasises the work of the project in the context of emerging digital library information environments, explores how a range of new linking tools might be combined and identifies ways in which different linking applications might converge. Some early results of linked pages from OpCit project are reported.

Although the OAI-PMH specification is focused on making it straightforward for data providers to expose metadata, practice shows that in certain significant situations deployment of OAI-PMH conformant repository software remains problematic. In this paper, we report on research aimed at devising solutions to further lower the barrier to make metadata collections harvestable. We provide an in depth description of an approach in which a data provider makes a metadata collection available as an XML file with a specific format an OAI Static Repository which is made OAI-PMH harvestable through the intermediation of software an OAI Static Repository Gateway - operated by a third party. We describe the properties of both components, and provide insights in our experience with an experimental implementation of a Gateway.

C-TORI (Cooperative TORI), a cooperative version of TORI (Task-Oriented Database Retrieval Interface), is presented in this paper. It extends interactive query formulation and result browsing by supporting cooperation between multiple users. In the cooperative environment, three basic additional operations are provided: copying, merging and coupling for three types of TORI objects (query forms, result forms, and query history windows). Cooperation with query forms allows end users to jointly formulate queries; cooperation with result forms supports users in jointly browsing through results and in sharing retrieved data without re-accessing the database; cooperative use of query histories yields a specific mechanism to share memory between users. The implementation is based on the concept of shared UI objects as an application-independent cooperation and communication model.

In this paper, we propose a method for the structural analysis of Web sites.The Web has become one of the most widely used media for electronic information because of its great flexibility. However, this flexibility has led to complicated structures. A structure that differs from the typical structures in a Web site might confuse readers, thus reducing the effectiveness of the site. A method for detecting unusual structures would be useful for identifying such structures so that their impact can be studied and ways to improve Web site effectiveness developed.We viewed the Web as a directed graph, and introduced a higher-order rank based on the non-well-founded set theory. We then developed higher-order rank analysis for detecting irregularities, defined as structures which differ from the typical structure of a target site. To test the effectiveness of our method, we applied it to several Web sites in actual use, and succeeded in identifying irregular structures in the sites.

Format: HTML Document(28K) .

Audience: HCI people, librarians. References: 16. Links: 3. Relevance: Low-Medium. Abstract: Starts off with a description of Berkeley's NSF/ARPA/NASA project, focusing on environmental data, particularly water planning for California. Diverse data types, bitmapped pages with OCR text. Describes the proper

ties of the task, and the users of the system. Talks about methods of assessing users needs, like interviews, observation, focus groups, etc. Claims that most users' expectations are too low, so user input doesn't provide appropriate

goals.

Format: HTML Document().

All knowledge work is, in some sense, collaborative. Trust is a key element of knowledge work: what we know depends largely on others. A better understanding of the epistemic machineries of knowledge communities and especially their practices of trust would be useful for designing effective digital libraries. This paper discusses the concepts of communities of practice and epistemic cultures, and their implication for design of digital libraries that support data sharing, with particular reference to practices of trust and credibility. It uses an empirical study of a biodiversity data system that collects and distributes data from a variety of sources to illustrate the implications of these concepts of knowledge communities for digital library design and operation. It concludes that diversity and uncomfortable boundary areas typify, not only digital library user groups, but the design and operation of digital libraries.

This paper describes several challenges that arise in designing a digital library for K12 education audiences when using Learning Object Metadata standard. These problems were multiplied when attempting to catalog the wide variety of informal learning and teaching resources from our museum's ever growing website and exhibit-based resource collections. This paper shares key challenges and early solutions for the creation of an educational metadata scheme based upon LOM, new vocabularies, and strategies for retrofitting existing informal learning science resources into learning objects.

We describe an implementation that has users `ck' notes, images, audio, and videoles into piles beyond the screen of the PDA. This scheme allows the PDA user to keep information close at hand without sacrcing valuable screen real estate. It also obviates the need to browse complexle trees during a working session. Multiple workspaces can be maintained in persistent store. Each workspace preserves one coguration of off-screen piles. The system allows multiple PDA owners within ad hoc radio range to share off-screen piles. They point out to each other where a shared pile is to reside in space. Once established, all sharing partners may add to the pile and see its contents. One application is to support biodiversity researchers in theeld, where they generate data on their PDA and need to keep it organized until they return to theireld station. We conducted an experiment where participants used our system with up to ten simultaneous piles. Not only were they able to operate the application, but they remembered the location of piles when placed in different physical environments and when asked to recall the locations several days after the experiment. We describe gender differences that suggest particular design choices for the system

MPEG is undertaking a new initiative to standardize content description of audio and video data/documents. When it is finalized in 2001, MPEG-7 is expected to provide standardized description language and schemes for concise and unambiguous content description of data/documents of complex media types. Meanwhile, other meta-data or description schemes, such as Dublin Core, XML, RDF, etc., are becoming popular in different application domains. In this paper, we propose Multimedia Description Framework (MDF), which is designated to accommodate multiple description (meta-data) schemes, MPEG-7 and non-MPEG-7, into integrated architecture. We will u`se examples to show how MDF description makes use of combined strength of different description schemes to enhance its expression power and flexibility. We conclude the paper with discussion of using MDF description of MPEG-7 Content Set to search/retrieve required audio and video documents from the set utilizing an MDF prototype system we have implemented.

Query-by-humming systems search a database of music for good matches to a sung, hummed, or whistled melody. Errors in transcription and variations in pitch and tempo can cause substantial mismatch between queries and targets. Thus, algorithms for measuring melodic similarity in query-by- humming systems should be robust. We compare several variations of search algorithms in an effort to improve search precision. In particular, we describe a new frame-based algorithm that significantly outperforms note-by-note algorithms in tests using sung queries and a database of MIDI-encoded music. Keywords dynamic programming, melodic comparison, melodic searching, Music Information Retrieval (MIR), sung query

In this paper, we propose a machine learning approach to title extraction from general documents. By general documents, we mean documents that do not belong to any specific genre, including presentations, book chapters, technical papers, brochures, reports, and letters. Previously, methods were proposed mainly for title extraction from research papers. It was not clear whether it is possible to conduct automatic title extraction from general documents. As case study, we consider extraction from Office including Word and PowerPoint. In our approach, we annotate titles in sample documents (for Word and PowerPoint respectively) and take them as training data, train machine learning models, and perform title extraction using the trained models. Our method is unique in that we mainly utilize format information such as font size as features in the models. It turns out that the use of format information is the key to a successful extraction from general documents. Precision and recall for title extraction from Word were 0.810 and 0.837 respectively, and precision and recall for title extraction from PowerPoint were 0.875 and 0.895 respectively in an experiment on intranet data. Other important new findings in this work include that we can train models in one domain and apply them to another domain, and more surprisingly we can even train models in one language and apply them to another language. Moreover, we can significantly improve search ranking results in document retrieval by using the extracted titles.

Research shows that recommendations comprise a valuable service for users of a digital library. While most existing recommender systems rely either on a purely content-based approach or purely collaborative approach to make recommendations, there is a need for digital libraries to use a combination of both approaches (a hybrid approach) to improve recommendations. In this paper, we report how we tested the idea of using a graph-based recommender system that naturally combines the content-based and collaborative approaches. Due to the similarity between our problem and a concept retrieval task, a Hopfield net algorithm was used to exploit high-degree book-book, user-user and book-user associations. Sample hold-out testing and preliminary subject testing were conducted to evaluate the system, by which it was found that the system gained improvement with respect to both precision and recall by combining content-based and collaborative approaches. But no significant improvement was observed by exploiting high-degree associations.

Recommender systems can provide valuable services in a digital library environment, as demonstrated by its commercial success in book, movie, and music industries. One of the most successful recommendation algorithms is collaborative filtering, which explores the correlations within user-item interactions to infer user interests and preferences. However, the recommendation quality of collaborative filtering approaches is greatly limited by the data sparsity problem. To alleviate this problem we have previously proposed graph-based algorithms that explore transitive user-item associations. In this paper, we extend from the idea of analyzing user-item interactions as graphs and employ link prediction approaches proposed in the recent network modeling literature for making collaborative filtering recommendations. We have adapted a wide range of linkage measures for making recommendations. Our preliminary experimental results based on a book recommendation dataset show that some of these measures achieved significantly better performance than standard collaborative filtering algorithms.

Format: Compressed PostScript().

Workflows are the structured activities that take place in information systems in typical business environments. These activities frequently involve several database systems, user interfaces, and application programs. Traditional database systems do not support workflows to any reasonable extent. Usually, human beings must intervene to ensure their proper execution. We have developed an architecture based on AI technology that automatically manages workflows. This architecture executes on top of a distributed computing environment. It has been applied to automating service provisioning workflows; an implementation that operates on one such workflow has been developed. This work advances the Camel Project's goal of developing technologies for integrating heterogeneous database systems. It is notable in its marriage of AI approaches with standard distributed database techniques.

Format: HTML Document().

This paper presents cross-language multilingual information retrieval using translated queries and a bilingual transfer dictionary. The experiments shows that multilingual IR is feasible, although performance lags considerably behind the monolingual standard.

This paper describes a Web-services-based system which we have developed to enable organizations to semi-automatically preserve their digital collections by dynamically discovering and invoking the most appropriate preservation service, as it is required. By periodically comparing preservation metadata for digital objects in a collection with a software version registry, potential object obsolescence can be detected and a notification message sent to the relevant agent. By making preservation software modules available as Web services and describing them semantically using a machine-processable ontology (OWL-S), the most appropriate preservation service(s) for each object can then be automatically discovered, composed and invoked by software agents (with optional human input at critical decision-making steps). We believe that this approach represents a significant advance towards providing a viable, cost-effective solution to the long term preservation of large-scale collections of digital objects.

Recent textbook on natural language translation.

This study analyzes metadata shared by cultural heritage institutions via the Open Archives Initiative Protocol for Metadata Harvesting. The syntax and semantics of metadata appearing in the Dublin Core fields creator, contributor, and date are examined. Preliminary conclusions are drawn regarding the effectiveness of Dublin Core in the Open Archives Initiative environment for cultural heritage materials.

The process of authoring document-centric XML documents in humanities disciplines is very different from the approach espoused by the standard XML editing software with the data-centric view of XML. Where data-centric XML is generated by first describing a tree structure of the encoding and then providing the content for the leaf elements, document-centric encodings start with content which is then marked up. In the paper we describe our approach to authoring document-centric XML documents and the tool, xTagger, originally developed for this purpose within the Electronic Boethius project, otherwise enhanced within the ARCHway project, an interdisciplinary project devoted to development of methods and software for preparation of image-based electronic editions of historic manuscripts.

Creating, maintaining, or using a digital library requires the manipulation of digital documents. Information workspaces provide a visual representation allowing users to collect, organize, annotate, and author information. The Visual Knowledge Builder (VKB) helps users access, collect, annotate, and combine materials from digital libraries and other sources into a personal information workspace. VKB has been enhanced to include direct search interfaces for NSDL and Google. Users create a visualization of search results while selecting and organizing materials for their current activity. Additionally, metadata applicators have been added to VKB. This interface allows the rapid addition of metadata to documents and aids the user in the extraction of existing metadata for application to other documents. A study was performed to compare the selection and organization of documents in VKB to the commonly used tools of a Web browser and a word processor. This study shows the value of visual workspaces for such effort but points to the need for sub-document level objects, ephemeral visualizations, and support for moving from visual representations to metadata.

Format: HTML Document().

SDARTS is a protocol and toolkit designed to facilitate metasearching. SDARTS combines two complementary existing protocols, SDLIP and STARTS, to define a uniform interface that collections should support for searching and exporting metasearch-related metadata. SDARTS also includes a toolkit with wrappers that are easily customized to make both local and remote document collections SDARTS-compliant. This paper describes two significant ways in which we have extended the SDARTS toolkit. First, we have added a tool that automatically builds rich content summaries for remote web collections by probing the collections with appropriate queries. These content summaries can then be used by a metasearcher to select over which collections to evaluate a given query. Second, we have enhanced the SDARTS toolkit so that all SDARTS-compliant collections export their metadata under the emerging Open Archives Initiative OAI) protocol. Conversely, the SDARTS toolkit now also allows all OAI-compliant collections to be made SDARTS- compliant with minimal effort. As a result, we implemented a bridge between SDARTS and OAI, which will facilitate easy interoperability among a potentially large number of collections. The SDARTS toolkit, with all related documentation and source code, is publicly available at http://sdarts.cs.columbia.edu.

We are creating an interactive tool to help non-professional web site builders create high quality designs. We have previously reported that quantitative measures of web page structure can predict whether a site will be highly or poorly rated by experts, with accuracies ranging from 67-80 several ways. First, we compute a much larger set of measures (157 versus 11), over a much larger collection of pages (5300 vs. 1900), achieving much higher overall accuracy (94 contrasting good, average, and poor pages. Second, we introduce new classes of measures that can make assessments at the site level and according to page type (home page, content page, etc.). Finally, we create statistical profiles of good sites, and apply them to an existing design, showing how that design can be changed to better match high-quality designs

We consider how to obtain a safe and efficient scheme for Web advertising. We introduce to cryptography the market model, a common concept from economics. This corresponds to an assumption of rational behavior of protocol participants. Making this assumption allows us to design schemes that are highly efficient in the common case - which is, when participants behave rationally. We demonstrate such a scheme for Web advertising. employing the concept of e-coupons. We prove that our proposed scheme is safe and meets our stringent security requirements.

Analyzes results from a user study related to the AHA (Audio HTML Access) framework, which tested three audio browsers to determine the appropriateness of certain types of audio markings for various HTML structures. The results added another dimension to the AHA framework, so that the principles outlined in it for choosing sounds to use in an audio presentation of HTML are now: (1) Vocal Source Identity (when to use speaker changes to mark structures), (2) Recognizability, and (3) Distraction (new)

Discusses application of the principles in the AHA framework to the actual choice of sounds in scenario interfaces. By looking at scenarios, we can see that other factors related to users (such as musical ability, culture, reading style, etc.) are needed in combination with the AHA principles to select specific sounds.

The Alexandria Digital Earth ProtoType (ADEPT) architecture is a framework for building distributed digital libraries of georeferenced information. An ADEPT system comprises of one or more autonomous libraries, each of which provides a uniform interface to one or more collections, each of which manages metadata for one or more items. The primary standard on which the architecture is based is the ADEPT bucket framework, which defines uniform client-level metadata query services that are compatible with heterogeneous underlying collections. ADEPT functionality strikes a balance between the simplicity of Web document delivery and the richness of Z39.50. The current ADEPT implementation runs as servlet-based middleware and supports collections housed in arbitrary relational databases.

To support non-trivial clients, such as data exploration and analysis environments, digital libraries must be able to describe the access modes that their contents support. We present a simple scheme that distinguishes four content accessibility classes: download (byte-stream retrieval), service (API), web interface (interactive), and alternative (semantically equivalent) or multipart (component) hierarchies. This scheme is simple enough to be easily supported by DL content providers, yet rich enough to allow programmatic clients to automatically identify appropriate access point(s).

The UpLib personal digital library system is specifically designed for secure use by a single individual. However, collaborative operation of multiple UpLib repositories is still possible. This paper describes two mechanisms that have been added to UpLib to facilitate community building around individual document collections.

Automatic metadata generation may provide a solution to the problem of inconsistent, unreliable metadata describing resources on the Web. The Resource Description Framework (RDF) provides a domain-neutral foundation on which extensible element sets can be defined and expressed in a standard notation. This paper describes how an automatic classifier, that classifies HTML documents according to Dewey Decimal Classification, can be used to extract context sensitive metadata which is then represented using RDF. The process of automatic classification is described and an appropriate metadata element set is identified comprising those elements that can be extracted during classification. An RDF data model and an RDF schema are defined representing the element set and the classifier is configured to output the elements in RDF syntax according to the defined schema.

Format: HTML Document (22K).

Audience: Publishers, slightly technical.. References: 0. Links: 0. Relevance: Low-medium. Abstract: A publisher's view on why publisher's won't be irrelevant. Also, a description of the type of security that a publisher would expect (ie, let's not worry about making it perfect, just reasonable.) -a header based system, without details of how to ensure the restrictions are obeyed. Gives a specific list of information that he thinks should be in the header.

This paper focuses on the multifaceted use of the OAI-PMH in a repository architecture designed to store digital assets at the Research Library of the Los Alamos National Laboratory (LANL), and to make the stored assets available in a uniform way to various downstream applications. In the architecture, the MPEG-21 Digital Item Declaration Language is used as the XML-based format to represent complex digital objects. Upon ingestion, these objects are stored in a multitude of autonomous OAI-PMH repositories. An OAI-PMH compliant Repository Index keeps track of the creation and location of all those repositories, whereas an Identifier Resolver keeps track of the location of individual objects. An OAI-PMH Federator is introduced as a single-point-of-access to downstream harvesters. It hides the complexity of the environment to those harvesters, and allows them to obtain transformations of stored objects. While the proposed architecture is described in the context of the LANL library, the paper will also touch on its more general applicability.

We describe WebWatcher as a tour guide agent for the web, the learning algorithms used by WebWatcher, experimental results based on learning from thousands of users, and lessons learned from this case study of tour guide agents.

Format: HTML Document(34K + pictures) .

Audience: Searchers, HCI people. References: 10. Links: 2. Relevance: Low. Abstract: Describes a MS Windows interface for searching using a thesaurus of related terms. Has 3 parts: a hierarchical organization of the terms, a cloud of related terms, and a keyword-in-context that tries to match what

you type incrementally. The cloud and hierarchy are point and click, and the hierarchy can be expanded and collapsed ala MS Word outline mode. Also capable of handling multi-hierarchies, where a term has multiple roots.

The Archive of the Indigenous Languages of Latin America (AILLA) is a web-accessible repository of multi-media resources in and about the indigenous languages of Latin America. In this paper, I describe the Graded Access System developed at AILLA to protect sensitive materials by allowing resource producers - academics and indigenous people - finely-grained control over the resources they house in the archive.

Soon many people will retrieve information from the Web using handheld, palmsized or even smaller computers. Although these computers have dramatically increased in sophistication, their display size is - and will remain - much smaller than their conventional, desktop counterparts. Currently, browsers for these devices present Web pages without taking account of the very different display capabilities. As part of a collaborative project with Reuters, we carried out a study into the usability impact of small displays for retrieval tasks. Users of the small screen were 50subjects. Small screen users used a very substantial number of scroll activities in attempting to complete the tasks. Our study also provided us with interesting insights into the shifts in approach users seem to make when using a small screen device for retrieval. These results suggest that the metaphors useful in a full screen desktop environment are not the most appropriate for the new devices. Design guidelines are discussed, here, Proposing directed access methods for effective small screen interaction. In our ongoing work, we are developing such 'meta-interfaces' which will sit between the small screen user and the `conventional' Web page.

Digital library research is made more robust and effective when end-user opinions and viewpoints inform the research, design and development process. A rich understanding of user tasks and contexts is especially necessary when investigation the use of mobile computers in traditional and digital library environments, since the nature and scope of the research questions at hand remain relatively undefined. This paper outlines findings from a library technologies user survey and on-site mobile library access prototype testing, and presents future research directions that can be derived from the results of these two studies.

Many digital libraries are comprised of documents from disparate sources that are independent of the rest of the collection in which they reside. A user's ability to explore is severely curtailed when each document stands in isolation; there is no way to navigate to other, related, documents, or even to tell if such documents exist. `We describe a method for automatically introducing topic-based links into documents to support browsing in digital libraries. Automatic keyphrase extraction is exploited to identify link anchors, and keyphrase-based similarity measures are used to select and rank destinations. Two implementations are described: one that applies these techniques to existing WWW-based digital library collections using standard HTML, and one that uses a wider range of interface techniques to provide more sophisticated linking capabilities. An evaluation shows that keyphrase-based similarity measures work as well as a popular full-text retrieval system for finding relevant destination documents.

This paper describes an evaluation of the Kea automatic keyphrase extraction algorithm. Tools that automatically identify keyphrases are desirale because document keyphrases have numerous applications in digital library sysems, but are costly and time consuming to manually assign. Keyphrase extraction algorithms are usually evaluated by comparison to author-specified keywords, but this methodology has several well-known shortcomings. The results presented in this pper are based on subjective evaluations of the quality and appropriateness of keyphrases by human assessors, and make a number of contributions. First they validate previous evaluations of Kea that rely on author keywords. Second, they show Kea's performance is comparable to that of similar systems that have been evaluated by human assessors. Finally, they justify the use of author keyphrases as a performance metric by showing that authors generally choose good keywords.

Format: HTML Document (32K) . Audience: Mostly non-technical, funders, Geographic IS people.