We examine one possible future for the library of a professional knowledge worker by examining the UpLib personal digital library system. The system consists of a full-text indexed repository accessed through an active agent via a Web interface. It is suitable for personal collections comprising tens of thousands of documents (including papers, books, photos, receipts, email, etc.), and provides for ease of document entry and access as well as high levels of security and privacy. Unlike many other systems of the sort, user access to the document collection is assured even if the UpLib system is unavailable. It is 'universal' in the sense that documents are canonically represented as projections into the text and image domains, and uses a predominantly visual user interface based on page images. UpLib can thus handle any document format which can be rendered as pages. Provision is made for alternative representations existing alongside the text-domain and image-domain representation, either stored or generated on demand. The system is highly extensible through user scripting, and is intended to be used as a platform for further work in document engineering.

CACHE is a system supporting collaborative analysis tasks. We describe CACHE and a study of the effects of its use on reducing confirmation bias in analysis—the tendency to seek information that confirms rather than refutes an initial belief. We compare collaborative vs individual (Nominal) use of CACHE, as well as groups made of members with diverse (Heterogeneous) beliefs vs Homogeneous beliefs. We found reduction in confirmation bias for Heterogeneous collaborative groups using CACHE and working alone, but Homogeneous groups accentuated initial biases, rather than debiasing. Benefits of CACHE, future extensions, and our perspective on design are discussed.

An intelligence analyst often needs to keep track of more facts than can be held in human memory. As a result, analysts use a notebook or evidence file to record facts learned so far. In practice, the evidence file is often an electronic document into which text snippets and hand-typed notes are placed. While this kind of evidence file is easy to read and edit, it provides little help for making sense of the captured information. We describe Entity Workspace, a tool designed to be used in place of a traditional evidence file. Entity Workspace combines user interface and entity extraction technologies to build up an explicit model of important entities (people, places, organizations, phone numbers, etc.) and their relationships. Using this model, it helps the analyst find and re-find facts rapidly, notice connections between entities, and identify good documents and entities to explore next.

In this short paper, we summarized recent work in the understanding of social information foraging, and models of how users socially foraging with diverse hints. In communities of practice that depend on foraging in overly rich information environments, there appears to be pressure to self-organize into a balance of some division of labor, plus some degree of cooperation. One way to bridge between different communities of users is to diversify their information sources. We have been examining the possibility of using a collaborative search engine to achieve this effect.

An experiment was conducted to assess the generality of navigation knowledge learned from interaction with the Degree-of-Interest (DOI) Tree. The degree of transfer of learning from the DOI Tree was compared to transfer of learning from a standard Internet Explorer browser. Twenty participants worked with either the DOI or Explorer browser in a Training Phase in which they performed information-finding or comparison tasks requiring navigation through a widely used Web directory. Degree of transfer was measured by effects on task completion time in a Transfer Phase in situations with varying degrees of similarity to the original training conditions. An analysis based on percent transfer scores indicated that participants trained on navigation of an information structure using the DOI showed superior transfer across a range of conditions as compared to participants trained using the Explorer.

Taking the form of physical books, virtual 3D books can be used as basic components of e-book systems, information workspaces, and digital libraries. This paper describes the page turning design of 3Book, a 3D book system that we recently developed. Our design aims to find a sensible balance among important factors such as visual realism, readability, interactivity, and scalability. To convey the impression of reading or viewing an actual physical book, we model all the faces of the book and synchronize the movements of various portions of the book during page turning. Our design delivers a seamless transition between two states of the book (i.e., when it is lying still and when it is turning pages). In addition, we deform the turning pages around an imaginary cone of changing sizes to produce realistically-looking curved pages.

Retyping text phrases can be time consuming. As a result, techniques for copying text from one software application to another, such as copy-and-paste and drag-and-drop are now commonplace. However, even these techniques can be too slow in situations where many phrases need to be copied. In the special case where the phrases to be copied represent syntactically identifiable entities, such as person names, company names, telephone numbers, or street addresses, much faster phrase copying is possible. We describe entity quick click, an approach that reduces both the amount of cursor travel and the number of button presses needed to copy a phrase.

We present an experiment that compares how people perform search tasks in a degree-of-interest browser and in a Windows-Explorer-like browser. Our results show that, whereas users do attend to more information in the DOI browser, they do not complete the task faster than in an Explorer-like browser. However, in both types of browser, users are faster to complete high information scent search tasks than low information scent tasks. We present an ACT-R computational model of the search task in the DOI browser. The model describes how a visual search strategy may combine with semantic aspects of processing, as captured by information scent. We also describe a way of automatically estimating information scent in an ontological hierarchy by querying a large corpus (in our case, Google's corpus).

In this paper we describe a comparison among three systems that estimate semantic similarity between words: Latent Semantic Analysis [6], Pointwise Mutual Information [17], and Generalized Latent Semantic Analysis [8]. We compare all these techniques on a unique corpus (TASA) and, for PMI and GLSA, we also report performance on a different web-based corpus. The evaluation is carried out through two kinds of tests: (1) synonymy tests, and (2) comparison with human word similarity judgments.

Methods for dimensionality reduction, notably LSI, have been successfully applied to the information retrieval task and document classification on small document collections. Since they involve a computation of the eigenvalue or singular value decomposition of a document-term matrix, their use for large real world applications is somewhat limited. In addition to it, the information about the term similarity that these methods can use is inferred only from the current document collection. We present an algorithm that computes a low dimensional vector space representation of documents using point-wise mutual information as a term similarity measure. Point-wise term similarity can be computed using any additional resources, such as the Web. Our method uses the term by term matrix and can therefore be applied to large document collections. Experimental results show a considerable performance improvement on the information retrieval tasks.

Methods for dimensionality reduction, notably LSA, have been successfully applied to the information retrieval task and document classification. Recently, corpus-based association measures such as point-wise mutual information have been found to outperform LSA on a variety of tasks. We have developed an algorithmic framework that computes a low-dimensional vector space representation of documents combining different measures of association with different dimensionality reduction techniques. Experimental results show a competitive performance on the synonymy and text classification tasks.

Document indexing and representation of term-document relations are very important issues for document clustering and retrieval. In this paper, we present Generalized Latent Semantic Analysis as a framework for computing semantically motivated term and document vectors. Our focus on term vectors is motivated by recent success of co-occurrence based measures of semantic similarity obtained from very large corpora. Our experiments demonstrate that GLSA term vectors efficiently capture semantic relations between terms and outperform related approaches on the synonymy test. We also show that termbased document representation improves performance on the document classification test.

Document indexing and representation of term document relations are very important issues for document clustering and retrieval. In this paper, we present Generalized Latent Semantic Analysis as a framework for computing semantically motivated term and document vectors. Our focus on term vectors is motivated by recent success of co-occurrence based measures of semantic similarity obtained from very large corpora. Our experiments demonstrate that GLSA term vectors efficiently capture semantic relations between terms and outperform related approaches on the synonymy test.

Any new system must provide significant advantages to users for them to adopt it over their existing practices. In this paper, we discuss changes made over the last two years of use of the UpLib personal digital library system, to provide those advantages in the realm of document management. These changes are concentrated in the document acquisition phase, where document analysis is performed and databases of document information are prepared. However, some changes have been made in the areas of document management and document usage, primarily to allow user better interaction with the improved document projections.

User interfaces for digital library systems must support a wide range of user activities. They include search, browsing, and curation, but perhaps the most important is actual reading of the items in the library. Support for reading, however, is usually relegated to applications which are only loosely integrated with the digital library system. One reason for this is the absence of toolkit widget support for the activity of reading. Most user interface toolkits instead provide support for either text editing or text presentation. This makes it difficult to write applications which support reading well. In this paper we describe the origins, design, and implementation of a new Java Swing toolkit widget called ReadUp, which provides support for reading page images in a digital library application, and discuss briefly how it is being used.

A digital library system must support a sweep of document activities for its users, including acquisition, organization, retrieval, and actual reading. In this paper, we discuss how these activities are supported at the base levels of the UpLib personal digital library system. We report on two new features of UpLib which augment previous capabilities. The UpLib Portal is a tool for document acquisition and subsequent retrieval. The ReadUp document reader is a component specifically designed to support reading with annotation and search.

Literature review and research roadmap for Department of Homeland Security National Visual Analytics program.

An advanced visual interface system is presented for fluid interaction in a personal digital library system. The system employs a zoomable planar representation of a collection using hybrid continuous/quantum treemap visualizations to facilitate navigation while minimizing cognitive load. By providing both fluidity and a means of reading documents within the same visualization, the system obliterates the traditional boundary separating the acquisition of materials from their use. In addition, the system provides a means of streamlining and largely automating the addition of new documents into a collection. The system is particularly well suited to user tasks which, in the physical world, are normally carried out by laying out a set of related documents on a physical desk — namely, those tasks that require frequent and rapid transfer of attention from one document in the collection to another. Discussed are the design and implementation of the system as well as its relationship to previous work.

The wide availability of digital reading material online is causing a major shift in everyday reading activities. Readers are increasingly skimming instead of reading in depth. Highlights are increasingly used in digital interfaces to direct attention toward relevant passages within texts. In this paper, we study the eye-tracking behavior of subjects using both keyword highlighting and a new highlighting technique called ScentHighlights, introduced recently [7]. In this first eye-tracking study of highlighting interfaces, we show that there is direct evidence of the von Restorff isolation effect [21] in the eye-tracking data, in that subjects performed better when a fact is isolated (highlighted) against a homogeneous background. Users with the ScentHighlights condition paid more attention to highlighted areas and are more accurate than with other interfaces. In addition to confirming the von Restorff effect, we found that there is great variation in subject differences in reading strategies among subjects, even in the presence of strong cues such as highlights. Some readers scan for highly profitable regions first, while others read sequentially despite the presence of strong highlight cues. The results point to future design possibilities in highlighting interfaces.

SNIF-ACT (Scent-based Navigation and Information Foraging in the ACT Cognitive architecture) is developed to explain navigation behavior on the World Wide Web when users are engaged in unfamiliar information-seeking tasks. SNIF-ACT makes navigation choices based on assessments of activation that spreads to the users' information goal from proximal cues (information scent) that occur in Web browsers. SNIF-ACT 1.0 was matched to the detailed protocol analyses from the N = 8 participants studied in Card, S., Pirolli, P., Van Der Wege, M., Morrison, J., Reeder, R., Schraedley, P., & Boshart, J. (2001). Information scent as a driver of Web Behavior Graphs: Results of a protocol analysis method for web usability. CHI 2001, ACM Conference on Human Factors in Computing Systems, CHI Letters, 3(1), 498-505. SNIF-ACT 1.0 was found to provide good fits to participants' choice of links to follow, and to their choice of when to leave a current patch of information. SNIF-ACT 2.0 incorporates several refinements based on a rational analysis of link selection on a web page. These refinements include a satisficing assessment of link selection from a Web page and an adaptive stopping rule that decides when to leave a Web page based on past and current experiences with a Web site. SNIF-ACT 2.0 was validated on a data set obtained from N = 244 participants studied by Chi, E. H., Rosien, A., Suppattanasiri, G., Williams, A., Royer, C., Chow, C., Robles, E., Dalal, B., Chen, J., & Cousins, S. (2003). The Bloodhound Project: Automating discovery of Web usability issues using the InfoScent simulator. CHI 2003, ACM Conference on Human Factors in Computing Systems, CHI Letters, 5(1), 505-512. Monte Carlo simulations of the model provided good fits to the Web navigation data. Theoretical and practical implications of the model are discussed.

No Abstract Available

The development of sports throughout history has been essential to our health and keeping us strong physically, mentally, and spiritually. Sport, as a complex human endeavor, promotes unity and brings the world together through the Olympic Games. As an intrinsic part of our social fabric, the various sports in the Olympics are fraught with rules and regulations, and, of course, controversy. The acceptance of technology in sports in the recent past is littered with controversy. Technological developments in sports equipment must therefore take into account of social acceptance issues, such as player perception of fairness and rule changes. In this article, we examine this issue from our experience of introducing a novel wearable sensor system for martial arts.

No Abstract Available

No Abstract Available

Within a more Information Foraging Theory, we have developed a rational analysis of Web use, which has shaped a cognitive model of Web navigation called SNIF-ACT. An automated and practical method for initializing the model with requisite knowledge of information scent was developed based on Pointwise Mutual Information (PMI) computations from a local document corpus with a Web back-off. An automated Web usability tool called Bloodhound was developed that implements an algorithm that approximates the operation of the cognitive model. We report on succesful empirical tests of the SNIF-ACT cognitive mode, the PMI method, and Bloodhound.

No Abstract Available

No Abstract Available

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.

This paper presents icon abacus, a space-efficient technique for displaying document attributes by automatic positioning of document icons. It displays the value of an attribute by using position on a single axis, allowing the other axis to display different metadata simultaneously The layout is stable enough to support navigation using spatial memory.

We investigate how people process negation in semantically distorted and metaphoric sentences. We present three experiments in which participants judged the truth of a_rmative and negative sentences that were either literal or contained semantic illusions (Erickson & Mattson, 1981) or metaphors. In all experiments, negation increases processing times; although for semantic illusions, negation preserves the ordering of judgment times for literal and nonliteral sentences, for metaphors this ordering is reversed, with nonliteral negatives taking less time than literal negatives. This result presents evidence against the traditional Clark and Chase (1972) model of negation. We propose the negation-as distortion model and discuss how this model can explain the data.

No Abstract Available

No Abstract Available

An advanced visual interface system is presented for fluid interaction in a personal digital library system. The system employs a zoomable planar representation of a collection using hybrid continuous/quantum treemap visualizations to facilitate navigation while minimizing cognitive load. By providing both fluidity and a means of reading documents within the same visualization, the system obliterates the traditional boundary separating the acquisition of materials from their use. In addition, the system provides a means of streamlining and largely automating the addition of new documents into a collection. The system is particularly well suited to user tasks which, in the physical world, are normally carried out by laying out a set of related documents on a physical desk - namely, those tasks that require frequent and rapid transfer of attention from one document in the collection to another. Discussed are the design and implementation of the system as well as its relationship to previous work.

Digital libraries are increasingly based on digital page images of documents, but techniques for constructing usable versions of these page images are still largely folklore. This paper documents some issues encountered in creating document icons, page thumbnails, and page images for the UpLib digital library system, and suggests answers for each of them, based on both problem analysis and user feedback. In particular, several algorithms for determining useful sizes of both document icons and large page images are discussed.

The UpLib personal digital library system is specifically designed for secure use by a single individual, instead of collaborative use by a number of people. 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.

Everyone engages in sense-making whether searching the Web, wading through boxes of documents or analyzing volumes of message traffic. Computers have a role to play, not just in processing ever-increasing amounts of information faster, but in augmenting our abilities to make sense of it all. Cognitive task analysis is the key to understanding sense-making activities and finding the technological leverage points.

Looking back on the tenth anniversary of the first Symposium on Information Visualization, much has been accomplished. The computer’s power has been exploited to give quick visual form to abstract data, to interact, and to warp detail to follow the user’s changing interest. Moreover, the design space of visualizations has been systematized with reference models, taxonomies, and monographs relating visualization to perceptual and graphical constraints. Looking ahead, however, I will argue that the era of pure information visualization is over. The path ahead depends on giving much more attention to the purposes of visualization and its use. Leaving aside communication, the purpose of information visualization is insight, or more particularly, a larger process that might be called sensemaking. I will sketch out the nature of sensemaking, exemplify it empirically in a practical, urgent setting, and suggest how theories of sensemaking could be developed. I will then describe systems that subsume information visualization as part of an emerging class of sensemaking systems combining visualization (the mind’s eye) with semantic content analysis and sensemaking operations (the mind’s muscle). Not surprisingly, a focus on sensemaking is a good generator of new visualizations. But these developments also suggest that it may be time for the information visualization field to alter its boundaries to go beyond the merely visual.

This paper extends previous work on focus+context visualizations of tree-structured data, introducing an efficient, space-constrained, multi-focal tree layout algorithm (“TreeBlock”) and techniques at both the system and interactive levels for dealing with scale. These contributions are realized in a new version of the Degree-Of-Interest Tree browser, supporting real-time interactive visualization and exploration of data sets containing on the order of a million nodes

Previous work on information foraging theory has addressed how people navigate through information systems to find information. This paper presents a new information foraging model called InfoCLASS that models the conceptual categories that people learn while interacting with information systems. InfoCLASS is based on previous work on the rational analysis of human category formation. InfoCLASS simulations can be used to make qualitative predictions about the richness of conceptual categories that will be learned from different experiential histories, and from different kinds of user interfaces. InfoCLASS simulations can also be used to assess the conceptual coherence among a group of users of information systems. It is argued that the degree of conceptual coherence among a group of users is an important determinant of the efficiency and effectiveness of a social organization engaged in making discoveries, such as scientific communities or intelligence agencies.

Presents rational analyses of information foraging on the Web

Summarizes the theory of information scent and related it to Egon Brunswik's probablistic functionalism

A number of real world applications require the evaluation and ranking of word pair association. The three most effective techniques for measuring word association are based on statistical techniques that use corpus co-occurrence counts, latent semantic analysis techniques, or a lexical ontology. A major limitation of latent semantic analysis and lexical ontology techniques is their inability to deal with out of corpus words. This limitation has motivated researchers to use the web as a corpus for computing co-occuranece counts. However, a significant number of real world applications require a large number of co-occuranece evaluations that cannot be readily computed from the Web due to search engine limitations on the number of queries. In this paper, we use a hybrid approach for computing world pair association. This approach uses statistical similarity measures computed from co-occurence counts in a local corpus in case of common terms or from the web in the case of infrequent terms. We compare the four models of semantic similarity. The models range in their access to syntactic and semantic informativeness as well as search engine access. We evaluate the models using human ratings of semantic similarity. The outcome shows that the best correlations with human similarity judgment comes from a statistical approach that uses a local corpus and backs off to the web. Experimental results on a number of evaluation data sets show a significant reduction in the number of search engine queries while maintaining the same performance levels. The storage and computational requirements of our hybrid approach can easily fit on a modern workstation.

Creativity often emerges out of the tension between two seemingly irreconcilable properties that a particular design or product should have. By creating a higher order synthesis that honors the essential elements, a creative solution arises. There is also a tension of opposites in methods for creative work: cultivating a “prepared mind” versus cultivating a “beginner’s mind.” Advice for the prepared mind says “Develop and use your experience” and advice for the beginner’s mind advice says “Discard your previous experience.” How can we effectively use these seemingly opposite pieces of advice?

This work compares user performance on a shape grouping task with a folder-based interface and a zooming interface. There was a significant difference in task completion times between the two interfaces, with subjects performing 30% faster with the zooming interface than with folders.

The Instant Bookplex (IB) system includes a zoomable user interface (ZUI) for navigating through a spatial representation of a document collection. This ZUI supports extended reading in the collection using semantic zooming, graphical presentation of metadata, animated transitions, and an integrated reading user interface. IB helps users find and re-find documents, choose good documents to read next, and navigate between documents.

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 of interest. However, we observe that while these tools provide significant help with relatively short-term reading projects involving small numbers of documents, they fall short of supporting readers who are engaged in longer-term reading projects, in which a topic is to be understood in-depth by reading many documents. These 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 a software tool designed to facilitate in-depth reading. This tool is a document collection browser that supports reading tasks that take place over days, weeks, or longer. The browser includes a zoomable user interface that helps the reader see the details of a document corpus in context. We describe the browser’s design and the design principles that motivated it.

Taking the form of physical books, virtual 3D books can be used as basic components of e-book systems, information workspaces, and digital libraries. This paper describes the page turning design of 3Book, a 3D book that we recently developed based on our previous experiences with WebBook. Our design achieves interactive page turning by employing page textures of multiple resolutions at different stages of page turning, solving the scalability problem. By modeling all the faces of the book and synchronizing the movements of various portions of the book during page turning, our design helps to convey the impression of reading or viewing an actual physical book.

We describe a novel technique for deforming the pages of virtual 3D books to produce a realistic page turning effect.

Ubiquitous and Wearable Computing both have the goal of pushing the computer into the background, supporting all kinds of human activities. Application areas include areas such as everyday environments (e.g. clothing, home, office), promoting new forms of creative learning via physical/virtual objects, and new tools for interactive design. In this paper, we thrust ubiquitous computing into the extremely hostile environment of the sparring ring of a martial art competition. Our system uses piezoelectric force sensors that transmit signals wirelessly to enable the detection of when a significant impact has been delivered to a competitor’s body. The objective is to support the judges in scoring the sparring matches accurately, while preserving the goal of merging and blending into the background of the activity. The system therefore must take into account of the rules of the game, be responsive in real-time asynchronously, and often cope with untrained operators of the system. We present a user study of the finished prototype and detail our experience.

The particular focus of this chapter will be on a psychological theory of information scent (Pirolli, 1997, 2003; Pirolli & Card, 1999) that is embedded in a broader model (Pirolli & Fu, 2003) of information foraging on the Web. The notion of information scent also has been used in developing models of people seeking information in document-clustering browsers (Pirolli, 1997) and highly interactive information visualizations (Pirolli, Card, & Van Der Wege, 2003). Information scent refers to the cues used by information foragers to make judgments related to the selection of information sources to pursue and consume. These cues include items such as Web links or bibliographic citations that provide users with concise information about content that is not immediately available. The information scent cues play an important role in guiding users to the information they seek, and they also play a role in providing users with an overall sense of the contents of collections. The purpose of this paper is to present a theoretical account of information scent that supports the development of models of navigation choice.

Although large displays are becoming more cost effective, most user interfaces are optimized for a single monitor of modest size even though many traditional workspaces such as desks and workbenches are much larger and some studies have found benefits from large displays. This paper explores whether a single monitor is sufficient for information work using standard software. A log-based longitudinal field study finds that most of the time a single monitor allows skilled information analysts to have a reasonable pattern of window activity. However, a novel visualization of the data shows that windows typically fill the monitor and the pattern is occasionally interrupted by window thrashing, the rapid manipulation of windows caused by limited display resource. Given these findings, we identify some common tasks that justify the development and the expense of wideband visual interfaces that are optimized for larger displays.

Our work in the past five years on modeling user actions on the Web has shown that a great deal of information about user actions can be recovered from the informational cues processed by the user during navigation. We call these informational cues by the name of "Information Scent." We have shown in various papers that Information Scent can be used as a methodology for clustering a group of user profiles [Chi02], simulating a collection of users navigating thru the Web with an information need [Chi03], and provid-ing navigational cues to users with transient information goals [Olston03]. We argue in this position paper for a CHI2004 workshop that more research in user profiling should be done for user goals that are transient in nature.

Wideband displays fill our field of view, creating new opportunities to develop effective visual interfaces. Although multiple monitors are becoming an affordable way to create wideband displays, the resulting seams create gaps in words and divide diagonal lines into nonaligned segments. We present several novel user interface techniques for creating seam-aware applications, proving that vendors need not wait for affordable seamless displays to exploit the potential of wideband displays.

This paper describes the 3Book, a 3D interactive visualization of a codex book as a component for various digital library and sensemaking systems. The book is designed to hold large books and to support sensemaking operations by readers. The book includes methods in which the automatic semantic analysis of the book’s content is used to dynamically tailor access.

This paper describes the 3Book, a 3D interactive visualization of a codex book as a component for digital library and information-intensive applications. The 3Book is able to represent books of almost unlimited length, allows users to read large format books, and has features to enhance reading and sensemaking.

Subject indexes were an important step forward for books because they enabled the comparison and correlations of information without extensive reading, re-reading and memorization. In this short paper, we focus on the user interaction and usage scenario of a new system called ScentIndex that enhances the subject index of an eBook by conceptually reorganizing it to suit particular information needs. Users first enter information needs via keywords describing the concepts they are trying to retrieve and comprehend. ScentIndex then computes what index entries are conceptually related, and reorganizes and displays these index entries on a single page.

We describe the design and use of a personal digital library system, UpLib. The system consists of a full-text indexed repository accessed through an active agent via a Web interface. It is suitable for personal collections comprising tens of thousands of documents (including papers, books, photos, receipts, email, etc.), and provides for ease of document entry and access as well as high levels of security and privacy. Unlike many other systems of the sort, user access to the document collection is assured even if the UpLib system is unavailable. It is ``universal'' in the sense that documents are canonically represented as projections into the text and image domains, and uses a predominantly visual user interface based on page images. UpLib can thus handle any document format which can be rendered as pages. Provision is made for alternative representations existing alongside the text-domain and image-domain representation, either stored or generated on demand. The system is highly extensible through user scripting, and is intended to be used as a platform for further work in document engineering. UpLib is assembled largely from open-source components (the current exception being the OCR engine, which is proprietary).

We trace through the initial design and early evolution of a personal digital library system pieced together largely from open-source components (the current exception being the OCR engine, which is proprietary). The system consists of a repository, a search engine, and a Web interface. It is suitable for personal collections comprising tens of thousands of documents (including papers, books, photos, receipts, etc.), and provides for ease of document entry and access as well as basic security and privacy. Representation of documents in an image-domain format is taken as canonical, exploiting the fact that documents are ultimately presented as visual objects to a user to achieve some degree of ``universality.'' Provision is made for alternative representations existing alongside the image-domain representation, either stored or generated on demand. At the conference we will demo the document-access portion of the system implemented self-contained on an ultra-portable laptop holding about 2,000 documents including many scanned books.

In the last 5 years, one major focus of the User Interface Research Group has been understanding how Information Foraging Theory could be used to explain user optimization behavior on the Web. My primary research focus has been on developing the Information Scent Model to describe local behaviors of a group of users around a Web locality. In order to validate this model, we have performed two major validation studies---one on the server end, and the other on the client end: -- We were interested in validating how Information Scent could be used to explain how users will move from page to page by following information cues that are present for each link, and evaluating the probabilities of finding the desired information by clicking through that link. We needed a large amount of user session data that is paired with user goals, and then simulating those goals to see if the user session data match. We performed this study _in situ_ with over 400 subjects at the client end using a remote user logging program, obtaining several thousand usable session data. -- We were also interested in how Information Scent could be used to perform more accurate Web usage data mining. We wanted to construct more accurate user profiles by applying Information Scent principles to Web usage log mining. We have obtained several large scale server logs, of which some sessions are associated with known user goals. We then built user profiles out of these logs. We measured the accuracy of our clustering algorithm by trying to match the user sessions with the known category of the user goals.

Information scent is a psychological theory of how people use perceptual cues, such as World Wide Web (WWW) links in order to make information-seeking decisions and to gain an overall sense of the contents of information collections. A spreading activation theory of information scent is presented, as well as a theory of learning categories (e.g., genres) of available information. The SNIF-ACT model incorporates information scent to model World Wide Web browsing. The InfoCLASS model simulates the learning of information categories from browsing.

Intelligence analysts (IAs) engage in information seeking, evaluation, prediction, and reporting behavior in an information-intensive work environment. A Cognitive Task Analysis (CTA) was conducted to capture data that will provide input to support development of a model of the analyst's processes, biases, and analytic strategies. A hybrid method was used to conduct the CTA, including a modified version of the critical decision method - the critical analysis method. The essential distinction was that participants were asked to describe an example of a strategic analysis problem. Procedures used to conduct the critical analysis method are described in this paper. Several factors contribute to making the IA's task challenging: (i) time pressure to produce reports in a shorter timeframe, (ii) a high cognitive workload, and (iii) difficult human judgments that are required regarding uncertain validity and reliability of the data. Human judgments are involved in considering the plausibility of information, deciding what information to trust, and determining how much weight to place on specific pieces of data. Intelligence analysis involves a complex process of assessing the reliability of information from a wide variety of sources and combining seemingly unrelated events. This problem is challenging because it involves aspects of data mining, data correlation and human judgment.

We present a new technique for efficiently computing Degree-of-Interest distributions to inform the visualization of graph-structured data. The technique is independent of the interest distribution used, and enables fluid interaction with very large data sets (over 100,000 nodes).

Although vendors have made multiple-monitor systems for many years, our interfaces have been stuck in a 30-year old windows paradigm focused on displays much smaller than the desktops we use when working with paper. Advances in flat panel displays and graphics cards now enable affordable personal computers with 6-8 monitors and may someday eliminate seams. This paper argues that vendors should be developing wideband visual interfaces that are designed for displays that fill the human visual field. We describe a longitudinal field study of window activity that found that windows almost always filled a typical single monitor display and that subjects occasionally struggled with window thrashing when they needed to work with two or more windows at the same time. Vendors need not wait for affordable seamless wideband displays before addressing these findings. We have implemented several novel user interface techniques for creating seam-aware applications that target wideband displays based on multiple monitors.

In this paper we describe a visualization architecture (AVID) that employs a dynamic model of user interest to support the design and creation of highly responsive, scalable visualizations of hierarchical data. We present evidence of the architecture's efficacy, showcasing dynamic visualizations with near-immediate (<100ms) update times, even on structures of over 100,000 nodes. We discuss how the key concepts used generalize to arbitrary graph structures. Additionally, we present the results of a user study comparing a prototypical visualization built using AVID to a more traditional file-browser interface, showcasing up to 20% improvement in information access times.

The two predominant paradigms for finding information on the Web are browsing and keyword searching. While they exhibit complementary advantages, neither paradigm alone is adequate for complex information goals that lend themselves partially to browsing and partially to searching. To integrate browsing and searching smoothly into a single interface, we introduce a novel approach called ScentTrails. Based on the concept of information scent developed in the context of information foraging theory, ScentTrails highlights hyperlinks to indicate paths to search results. This interface enables users to interpolate smoothly between searching and browsing to locate content matching complex information goals effectively. In a preliminary user study, ScentTrails enabled subjects to find information more quickly than by either searching or browsing alone.

No Abstract Available

No Abstract Available

According to usability experts, the top user issue for Web sites is difficult navigation. We have been developing auto-mated usability tools for several years, and here we describe a prototype service called InfoScent™ Bloodhound Simula-tor, a push-button navigation analysis system, which auto-matically analyzes the information cues on a Web site to produce a usability report. We further build upon previous algorithms to create a method called Information Scent Absorption Rate, which measures the navigability of a site by computing the probability of users reaching the desired destinations on the site. Lastly, we present a user study involving 244 subjects over 1385 user sessions that show how Bloodhound correlates with real users surfing for in-formation on four Web sites. The hope is that, by using a simulation of user surfing behavior, we can reduce the need for human labor during usability testing, thus dramatically lower testing costs, and ultimately improving user experience. The Bloodhound Project is unique in that we apply a concrete HCI theory directly to a real-world prob-lem. The lack of empirically validated HCI theoretical model has plagued the development of our field, and this is a step toward that direction.

Theoretical integration of cognitive models of memory, concept formation, Random Utility Model, and navigation

Theory for predicting user flow and inferring user goals on the WWW and similar systems

Briefly summarize the new/novel technical contribution: A thoery of how users process proximal cues to make navigation decision and develop conceptual representations of information sources.

No Abstract Available

A breakthrough creates something new or satisfies a previously undiscovered need. Radical breakthroughs often have uses and effects far beyond what their inventors had in mind and can launch new industries or transform existing ones. Our capacity for breakthroughs depends on a combination of science, imagination and business; the next waves of innovation will come from organizations that get this combination right. The stories in this book, told by repeat inventors and managers of technology, uncover the best practices for inventing the future.

This paper proposes Degree-of-Interest trees. These trees use degree-of-interest calculations and focus+context visualization methods, together with bounding constraints, to fit within pre-established bounds. The method is an instance of an emerging "attention-reactive" user interface whose components are designed to snap together in bounded spaces.

This demonstration shows a method and implementation to interactively display large hierarchies (up to 10,000 nodes) within a web browser. This software computes a degree of interest (DOI) for each node in the hierarchy and displays an overview of the complete hierarchy while showing more detail for nodes with a higher DOI value.

This paper proposes Degree-of-Interest trees. These trees use degree-of-interest calculations and focus+context visualization methods, together with bounding constraints, to fit within pre-established bounds. The method is an instance of an emerging “attention-reactive” user interface whose components are de-signed to snap together in bounded spaces.

Web Usage Mining enables new understanding of user goals on the Web. This understanding has broad applications, and traditional mining techniques such as association rules have been used in business applications. We have developed an automated method to directly infer the major groupings of user traffic on a Web site [Heer01]. We do this by utilizing multiple data features in a clustering analysis. We have performed an extensive, systematic evaluation of the proposed approach, and have discovered that certain clustering schemes can achieve categorization accuracies as high as 99% [Heer02b]. In this paper, we describe the further development of this work into a prototype service called LumberJack, a push-button analysis system that is both more automated and accurate than past systems.

Visualization can be viewed as a process that transforms raw data (value) into views. There has been two major category of data process models that have been proposed to model the visualization transformation process. This paper seeks to compare the Data Flow Models and the Data State Models. Specifically, it proves that, in terms of expressiveness, anything that can represented using the Data Flow Model can also be represented using the Data State Model, and vice versa.

No Abstract Available

No Abstract Available

In this paper, we introduce a concise and concrete definition of an accurate colon centerline and provide an efficient automatic means to extract the centerline and its associated branches (caused by a forceful touching of colon and small bowel or a deep fold in twisted colon lumen). We further discuss its applications on fly-through path planning and endoscopic simulation, as well as its potential to solve the challenging touching and colon collapse problems in virtual colonoscopy. Experimental results demonstrated its centeredness, robustness, and efficiency.

Predictive Web usage visualizations can help analysts uncover traffic patterns and usability problems.

Information visualization is the design and creation of interactive graphic depictions of information by combining principles in the disciplines of graphic design, cognitive science, and interactive computer graphics. This book describes a framework to make information visualization systems easier to develop through the creation of a reference model. It develops and discusses the general utility of this Data State Model, and validates it by applying it to various visualization techniques and showing several systems that illustrate issues such as how to model operators and interactions in visualization systems. The book also applies this reference model to make information visualization more accessible to potential users by creating a `Visualization Spreadsheet', where each cell can contain an entire set of data represented using interactive graphics.

This paper describes the visual analysis tool WebQuilt, a web usability logging and visualization system that helps web design teams record and analyze usability tests. The logging portion of WebQuilt unobtrusively gathers clickstream data as users complete specified tasks. This data is then aggregated and presented as an interactive graph, where nodes of the graph are images of the web pages visited, and arrows are the transitions between pages. To aid analysis of the gathered usability test data, the WebQuilt visualization provides filtering capabilities and semantic zooming, allowing the designer to understand the test results at the gestalt view of the entire graph, and then drill down to sub-paths and single pages. The visualization highlights important usability issues, such as pages where users spent a lot of time, pages where users get off track during the task, navigation patterns, and exit pages, all within the context of a specific task. WebQuilt is designed to conduct remote usability testing on a variety of Internet-enabled devices and provide a way to identify potential usability problems when the tester cannot be present to observe and record user actions.

Recent research has explored web user session clustering as a means of understanding user activity and interests on the World Wide Web. Though the proposed techniques have proven to be useful and effective, they require that one either specify the number of clusters in advance or browse a large hierarchy of clusters to find the optimal depth at which to describe user activity. In this paper, we examine the utility of a stability-based technique for automatically determining the optimal number of clusters in the context of web user session clustering. We present two case studies evaluating the technique’s effectiveness.

The Fluid Documents project has developed various research prototypes that show that powerful annotation techniques based on animated typographical changes can help readers utilize annotations more effectively. Our recently-developed Fluid Open Hypermedia prototype supports the authoring and browsing of fluid annotations on third-party Web pages. This prototype is an extension of the Arakne Environment, an open hypermedia application that can augment Web pages with externally stored hypermedia structures. This paper describes how various Web standards, including DOM, CSS, XLink, XPointer, and RDF, can be used and extended to support fluid annotations.

The Special Interest Group on Computer Human Interaction (SIGCHI) has had a successful history of 20 years of growth in its numbers and influence. To help guide the continued evolution of the academic discipline and professional community, we invite several senior members to offer their visions for what the field of CHI actually accomplished over the past several decades, and what do we still need to accomplish? What do we need to do differently/better/smarter? What haven't we tried because the technology, the money or the will wasn't there in the past, but perhaps is now? The CHI field is more than just technology. We understand that our work can have a profound effect on individuals, families, neighborhoods, corporations, and countries. We know that we can influence education, commerce, healthcare, and government. How can we contribute to bridging the digital divides in developed and developing countries? What agendas can we offer for the academic, research, industrial, and civic spheres for the next 20 years? How can we be more ambitious? How can we truly serve human needs?

On the Web, users typically forage for information by navigating from page to page along Web links. Their surfing patterns or actions are guided by their information needs. Researchers need tools to explore the complex interactions between user needs, user actions, and the structures and contents of the Web. In this paper, we describe two computational methods for understanding the relationship between user needs and user actions. First, for a particular pattern of surfing, we seek to infer the associated information need. Second, given an information need, and some pages as starting points, we attempt to predict the expected surfing patterns. The algorithms use a concept called “information scent”, which is the subjective sense of value and cost of accessing a page based on perceptual cues. We present an empirical evaluation of these two algorithms, and show their effectiveness.

Virtual endoscopy is a computerized, noninvasive procedure for detecting anomalies inside human organs. Several preliminary studies have demonstrated the benefits and effectiveness of this modality. Unfortunately, previous work cannot guarantee that an existing anomaly will be detected, especially for complex organs with multiple branches. In this paper, we introduce the concept of reliable navigation, which ensures the interior organ surface is fully examined by the physician performing the virtual endoscopy procedure. To achieve this, we propose computing a reliable fly-through path that ensures no blind area during the navigation. Theoretically, we discuss the criteria of evaluating a reliable path and prove that the problem of generating an optimal reliable path for virtual endoscopy is NP-complete. In practice, we develop an efficient method for the calculation of an effective reliable path. First, a small set of center observation points are automatically located inside the hollow organ. For each observation point, there exists at least one patch of interior surface visible to it, but that cannot be seen from any of the other observation points. These chosen points are then linked with a path that stays in the center of the organ. Finally, new points inside the organ are recursively selected and connected into the path until the entire organ surface is visible from the path. We present encouraging results from experiments on several data sets. For a medium size volumetric model with several hundred thousand inner voxels, an effective reliable path can be generated in several minutes.

Understanding user behaviors on Web sites enables site owners to make sites more usable, ultimately helping users to achieve their goals more quickly. Accordingly, researchers have devised methods for categorizing user sessions in hopes of revealing user interests. These techniques build user profiles by combining users' navigation paths with other data features, such as page viewing time, hyperlink structure, and page content. Previously, we have presented complex techniques of combining many of these data features to cluster user profiles. In this paper, we introduce a user study and a systematic evaluation of these different data features and their associated weighting schemes. We present the results of our study, including accuracy measures for a number of clustering approaches, and offer recommendations for Web analysts. While further investigation over more sites is needed to definitively settle on a robust scheme, we have characterized this analytic space.

WebQuilt is a web logging and visualization system that helps web design teams run usability tests (both local and remote) and analyze the collected data. Logging is done through a proxy, overcoming many of the problems with server-side and client-side logging. Captured usage traces can be aggregated and visualized in a zooming interface that shows the web pages people viewed. The visualization also shows the most common paths taken through the web site for a given task, as well as the optimal path for that task, as designated by the designer. This paper discusses the architecture of WebQuilt and also describes how it can be extended for new kinds of analyses and visualizations.

No Abstract Available

On the Web, users typically forage for information by navigating from page to page along Web links. Their surfing patterns or actions are guided by their information needs. Researchers need tools to explore the complex interactions between user needs, user actions, and the structures and contents of the Web. In this paper, we describe two computational methods for understanding the relationship between user needs and user actions. First, for a particular pattern of surfing, we seek to infer the associated information need. Second, given an information need, and some pages as starting points, we attempt to predict the expected surfing patterns. The algorithms use a concept called “information scent”, which is the subjective sense of value and cost of accessing a page based on perceptual cues. We present an empirical evaluation of these two algorithms, and show their effectiveness.

Eye trackers output a stream of points at which the eye was looking. To give these points meaning, researchers analyzing eye tracking data need to map the points onto the meaningful objects at which the eye was looking. Performing this mapping has proven to be a tedious, time-consuming task. We present a software system that automates this task for Web usability studies that incorporate eye tracking.

We introduce a technique for creating novel, textually-enhanced thumbnails of Web pages. These thumbnails combine the advantages of image thumbnails and text summaries to provide consistent performance on a variety of tasks. We conducted a study in which participants used three different types of summaries (enhanced thumbnails, plain thumbnails, and text summaries) to search Web pages to find several different types of information. Participants took an average of 67, 86, and 95 seconds to find the answer with enhanced thumbnails, plain thumbnails, and text summaries, respectively. We found a strong effect of question category. For some questions, text outperformed plain thumbnails, while for other questions, plain thumbnails outperformed text. Enhanced thumbnails (which combine the features of text summaries and plain thumbnails) were more consistent than either text summaries or plain thumbnails, having for all categories the best performance or performance that was statistically indistinguishable from the best.

In this paper, we present three taxonomic classification schemes based on Web users' responses to what Web activities significantly impacted their decisions and actions. The taxonomic classifications focus on three variables: the Purpose of people's search on the Web, the Method people use to find information, and the Content of the information for which they are searching. These taxonomies are useful for understanding people's activity on the Web and for developing ecologically-valid tasks to be used when studying web behavior.

In this paper, we describe an electronic guidebook prototype and report on a study of its use in a historic house. Supported by mechanisms in the guidebook, visitors constructed experiences that had a high degree of interaction with three entities: the guidebook, their companions, and the house and its contents. For example, we found that most visitors played audio descriptions through speakers (rather than using headphones or reading textual descriptions) to facilitate communication with their companions.

We describe tap tips, a technique for providing touch-screen target location hints. Tap tips are lightweight in that they are non-modal, appear only when needed, require a minimal number of user gestures, and do not add to the standard touchscreen gesture vocabulary. We discuss our implementation of tap tips in an electronic guidebook system and some usability test results.

Eye tracking studies of the Hyperbolic Tree browser suggest that visual search in focus+context displays is highly affected by information scent (i.e., local cues, such as text summaries, used to assess and navigate towards distal information sources). When users detected a strong information scent, they were able to reach their goal faster with the Hyperbolic Tree browser than with a conventional browser. When users detected a weak scent or no scent, users exhibited less efficient search of areas with a high density of visual items. In order to interpret these results we present an integration of the CODE Theory of Visual Attention (CTVA) with information foraging theory. Development of the CTVA-foraging theory could lead to deeper analysis of interaction with visual displays of content, such as the World Wide Web or information visualizations.

The purpose of this paper is to introduce a replicable WWW protocol analysis methodology illustrated by application to data collected in the laboratory. The methodology uses instrumentation to obtain detailed recordings of user actions with a browser, caches Web pages encountered, and videotapes talk-aloud protocols. We apply the current form of the method to the analysis of eight Web protocols, visualizing the structure of the interaction and showing the strong effect of information scent in determining the path followed.

The complexity of software has continuously risen since the invention of the computer, and while Moore's law predicts the growth in processor speed, it fails to take into account our ability in managing complex software processes. Our utilization of the increase in processor speed is very much dependent on our ability to manage this complexity.

Previously, we presented a system called AlignmentViewer that uses information visualization techniques to visualize similarities between a single DNA sequence and a large database of other sequences. In this paper, we extend, summarize, and describe the system using several interesting case studies. We present our comb glyph technique for visualizing alignments between sequences. In this paper, we also extend the original system by incorporating computational steering, and the visualization of differences between data sets. The case studies and the new extended system present our novel approach of extracting significant relationships in the biological data set.

Data visualization environments help users understand and analyze their data by permitting interactive browsing of graphical representations of the data. To further facilitate understanding and analysis, many visualization environments have special features known as portals, which are sub-windows of a data canvas. Portals provide a way to display multiple graphical representations simultaneously, in a nested fashion. This makes portals an extremely powerful and flexible paradigm for data visualization. Unfortunately, with this flexibility comes complexity. There are over a hundred possible ways each portal can be configured to exhibit different behaviors. Many of these behaviors are confusing and certain behaviors can be inappropriate for a particular setting. It is desirable to eliminate confusing and inappropriate behaviors. In this paper, we construct a taxonomy of portal behaviors and give recommendations to help designers of visualization systems decide which behaviors are intuitive and appropriate for a particular setting. We apply these recommendations to an example setting that is fully visually programmable and analyze the resulting reduced set of behaviors. Finally, we consider a real visualization environment and demonstrate some problems associated with behaviors that do not follow our recommendations.

In previous work, researchers have attempted to construct taxonomies of information visualization techniques by examining the data domains that are compatible with these techniques. This is useful because implementers can quickly identify various techniques that can be applied to their domain of interest. However, these taxonomies do not help the implementers understand how to apply and implement these techniques. In this paper, we will extend and then propose a new way to taxonomize information visualization techniques by using the Data State Model. In fact, as the taxonomic analysis in this paper will show, many of the techniques share similar operating steps that can easily be reused. The paper shows that the Data State Model not only helps researchers understand the space of design, but also helps implementers understand how information visualization techniques can be applied more broadly.

Item selection is a key problem in electronic guidebook design. Many systems do not apply so-called "context-awareness" technologies to infer user interest, placing the entire burden of selection on the user. Conversely, to make selection easier, many systems automatically eliminate information that they infer is not of interest to the user. However, such systems often eliminate too much information, preventing the user from finding what they want. To realize the full potential of electronic guidebooks, designers must strike the right balance between automatic context-based inference and manual selection. In this paper, we introduce a task-oriented model of item selection for electronic guidebooks to help designers explore this continuum. We argue that item selection contains three sub-tasks and that these sub-tasks should be considered explicitly in system design. We apply our model to existing systems, demonstrating pitfalls of combining sub-tasks, and discuss how our model has improved the design of our own guidebook prototype.

No Abstract Available

Considering the amount of interest in studying Web-browsing behavior, there is a relative lack of tools for data collection in this area. Those tools that do exist have significant limitations on the data they are able to collect or on their suitability for efficient analysis. We present WebLogger, a tool which instruments Microsoft's Internet Explorer Web browser. We have found that WebLogger alleviates some of the problems associated with other approaches to browser-based data collection methods.

A multiple view system uses two or more distinct views to support the investigation of a single conceptual entity. Many such systems exist, ranging from computer-aided design (CAD) systems for chip design that display both the logical structure and the actual geometry of the integrated circuit to overview-plus-detail systems that show both an overview for context and a zoomed-in-view for detail. Designers of these systems must make a variety of design decisions, ranging from determining layout to constructing sophisticated coordination mechanisms. Surprisingly, little work has been done to characterize these systems or to express guidelines for their design. Based on a workshop discussion of multiple views, and based on our own design and implementation experience with these systems, we present eight guidelines for the design of multiple view systems.

Focus + context information visualizations have sought to amplify human cognition by increasing the amount of information immediately available to the user. We study how the focus + context distortion of the Hyperbolic Tree browser affects information foraging behavior in a task similar to the CHI '97 Browse Off. In comparison to a more conventional browser, Hyperbolic users searched more nodes, searched at a faster rate, and showed more learning. However, the performance of the Hyperbolic was found to be highly affected by "information scent", proximal cues to the value of distal information. Strong information scent made hyperbolic search faster than with a conventional browser. Conversely, weak scent put the hyperbolic tree at a disadvantage. There appears to be two countervailing processes affecting visual attention in these displays: strong information scent expands the spotlight of attention whereas crowding of targets in the compressed region of the Hyperbolic narrows it. The results suggest design improvements.

Visual computational steering environments extend traditional visualization environments by enabling the user to interactively steer the computations applied to the data. In this paper, we develop a new type of computational steering. "Resource steering" extends current visual steering techniques by providing machine resource estimatio