Presented during the VisWeek 2010 Doctoral Colloquium.

Information Visualization for Web-based Search and ExplorationMarian Dork∗

University of Calgary

ABSTRACT

During my doctoral research I investigate how visual exploration af-fects information seeking on the Web. I am particularly interested inthe role of interactive visualization in exploratory search interfaces,development of web-based systems, and new ways of informationseeking. In this research overview, I provide an introduction to myresearch topic, outline prior related work, discuss my methodologyand challenges, and present a timeline for my research progress.

1 INTRODUCTION

With the exponential growth of the World Wide Web since its birth20 years ago, we experience an unparalleled information explosionthat raises many questions for research and ideas for innovation. Itcan be argued that the Web yields social and technological trans-formations only comparable to the modern printing press inventedmore than half a millennium ago. Both inventions have triggeredstrikingly similar reactions. Scholars of early modernity becameworried about keeping up with the rapid publication of books, yetat the same time they were also intrigued by the opportunity ofgrowing information access [21]. Similarly, the overabundance ofdocuments, media, and data on the Web can be viewed through thelenses of information overload as well as greater information ac-cess. With my doctoral research I am addressing both the challengeand opportunity of growing information spaces on the Web.

Problem The Web began as an information space dominatedby basic hypertext. Since then, it has also become a platform for so-phisticated interactivity and increasingly advanced graphics. At thesame time, it features rich links between diverse resources indicat-ing social, spatial, and semantic relationships. These advances havegradually expanded the range of available resources on the Web,however, with regard to the interaction techniques and user experi-ence web-based information seeking has not changed significantly.The canonical interface is a query box, followed by search resultstypically displayed as a long textual list also known as “10 bluelinks”. Conventional search is often used for exploratory tasks bychanging search queries and browsing different result sets. Whilethis approach typically leads to some relevant resources, it can bedifficult to develop an overview of the whole information space andhave a sense of orientation within it.

Objectives The primary goal of my research is to develop acomprehensive understanding of how interactive visualization cansupport information seekers in exploring information spaces on theWeb. Information visualization is known to help the viewer to seeoverviews, relationships, and patterns in abstract data [6]. With myresearch I am investigating how information visualization can sup-port searchers as they engage in exploratory search tasks with vagueand changing information needs. In particular, I am interested inhow visualization can affect the searcher’s sense of overview, cu-riosity, and serendipity during information seeking. I design visualinterfaces and study how they can help to orient oneself in an infor-mation space, remain interested and motivated, and discover new

∗http://mariandoerk.de

and possibly unexpected information. Rather than merely enter-ing text queries and browsing through result sets, interactive visu-alizations may provide a more effective way to formulate complexqueries and situate resources in the context of their collection.

Research Questions The fundamental hypothesis of my re-search is that visual exploration is a novel type of information seek-ing that can complement more established web-based interactiontechniques, such as browsing and searching, as a more high-levelway of navigating large volumes of resources. To test this hypothe-sis I pose the following research questions:

Q1: How can web resources be represented and explored with in-formation visualization?

Q2: How can visual exploration systems be architected and real-ized on the Web?

Q3: What are the effects of visualization on information seeking?

Scope My research falls into the cross section between threemain research communities that are associated with my researchquestions: information visualization (Q1), web research (Q2), andinformation seeking (Q3). The pairwise cross sections betweenthese communities represent currently active research areas: web-based information seeking, visualization for information seeking,and information visualization on the Web. At the overlap of theseresearch communities and areas, I situate my doctoral research onvisualization for web-based search and exploration (see Figure 1).

Information Visual-ization on the Web

Visualization forInformation Seeking

Web-based Information Seeking

Visualization forWeb-based Search

and Exploration

InformationSeeking

WebResearch

InformationVisualization

Figure 1: My PhD research lies at the cross section between infor-mation visualization, web research, and information seeking.

In the remainder of this research overview, I will first give a sum-mary of prior work most relevant for my research (Section 2). ThenI will briefly discuss my methodology (Section 3) and outline themain challenges for my doctoral research (Section 4). I will pro-vide a timeline of accomplished and planned projects and researchphases (Section 5), before I close with a conclusion.

2 BACKGROUND

My research builds on previous work in interdisciplinary fields de-voted to the study of information seeking and visualization in thecontext of digital information spaces, in particular the Web.

1

Presented during the VisWeek 2010 Doctoral Colloquium.

2.1 Web-based Information Seeking

Traditional information retrieval frames information seeking interms of search queries and the corresponding set of results max-imizing precision and recall, e. g. [2]. In contrast, there has beenmuch research recently that focuses on the searcher’s motivationsand needs. For example, Bates characterizes online informationseeking as evolving berrypicking that involves several informationsources and search tools [3]. She described the process of search-ing for digital information as multiple successive search queriesthrough an information space while their information need dynam-ically changes. The searcher is engaging in a form of informa-tion foraging weighing the cost of retrieving a resource againstits information value [17]. Similar to these ‘food gathering’ mod-els, Kuhlthau describes information seeking as a dynamic process,throughout which the information seeker’s feelings, thoughts, andactions change depending on the phase they are in [14].

Information seeking in hypertext environments has been dis-cussed as the dyad of searching and browsing [3]. These typesof information seeking typically require the formulation of mul-tiple queries, and consideration of many individual items, whichbecomes problematic with increasing numbers of informationsources. There is growing interest in going beyond the ‘10 bluelinks’ [5] and support a more exploratory search [26], but few novelinteraction techniques have been proposed so far. Faceted naviga-tion can be seen as an exploratory search technique providing thesearcher with multiple explorable categorizations that can be navi-gated without entering explicit search queries [10]. While complexfaceted navigation interfaces can lead to cognitive overload [27],first efforts are made to enhance them with visualization [23].

2.2 Visualization for Information Seeking

Much research was done on visualization for information seeking.For example, three-dimensional interfaces were proposed for fo-cus+context views [16], hierarchy visualizations [20], and windowlayouts [7]. These systems feature views representing informationstructures, such as filesystems and hypertext links, however, the ac-tual content of a collection is seldom represented. Exploring the useof visualizations for querying databases, interactive graphical ele-ments can act as visual filters coupled with visualizations [1] andquery terms can be interactively visualized [13, 22].

Search results can also be visualized helping searchers assess re-trieved documents. Per-result visualizations can, for example, rep-resent document-query similarities integrated into result lists [12].Result thumbnails can be positioned according to a visualizationsuch as a bar chart, as demonstrated by the zoomable search in-terface Pivot1. Overview visualizations can provide context for re-trieved search results [8] or represent whole document collectionsusing spatialization methods [28], which is still computationally ex-pensive, in particular, for large information spaces, such as the Web.There has been some work on integrating faceted navigation andvisualization [15, 23], hinting at the contributions that I attempt tomake with my doctoral research.

2.3 Information Visualization on the Web

As the Web expands in terms of range and number of resources andpossibilities for interactive graphics, we witness more visualiza-tions running in web browsers and a visualization practice that tar-gets casual uses [18]. For example, baby name visualizations [25],communities centred around visualizations [24], and visualizationsof emotions expressed in social media [11] share an increased em-phasis on playfulness, community, and aesthetics. While client-sidesoftware libraries make it easier to bring interactive visualization tothe Web [4], it remains an open challenge to ensure responsive-

1 http://www.getpivot.com/

ness and flexibility when visualizing large and diverse informationspaces distributed over the Web.

Two visualization techniques that have become widely spreadacross the Web are tag clouds and geographic maps. Studies haveshown that tag clouds support a range of information seeking tasksthrough different ways of formatting and positioning text [19]. Withthe introduction of Google Maps and the web-based mapping API,web developers are able to embed rich interactive maps into theirapplications and represent their resources geospatially.

3 METHODOLOGY

With aforementioned research questions and prior related work inmind, I am developing a hybrid methodology for my PhD researchthat integrates design, engineering, and evaluation and allows forboth quantitative and qualitative data collection.

I am undertaking several integrated case studies that combine vi-sualization, interface, and system design with user studies and per-formance evaluation. I am designing novel visualizations and in-teraction techniques for exploring information in large web-basedcollections and develop flexible systems that can be re-used andadapted for different types of information spaces. Based on theexperience and insight gained throughout the case studies, I willformulate a framework that integrates empirical insights and practi-cal design guidelines for web-based visual information explorationsystems. I am also interested in exploring the limitations of inter-active visualizations for different types of information needs.

My methodology differs from research undertaken in lab set-tings, as I make each prototype available on the Web, instrumentedfor capturing anonymized logs in order to study both usage andperformance. Web-based prototypes enable me to study interactionstrategies of many participants. For example, a web-based study Iundertook as part of my PhD research attracted over 750 partici-pants. Such high numbers of participants allow me to run compar-ative studies contrasting, for example, visual exploration with con-ventional text-based searching, to identify strengths and limitationsof both. However, a purely web-based methodology might sacrificesome of the descriptive detail that co-located observation and open-ended interviews give. To counter this I will undertake qualitative,observational studies as part of the case studies and I am develop-ing novel methods for web-based research that go beyond simplisticserver logs. For example, it is possible to do light-weight, client-side cursor tracking to generate heat maps representing interfaceuse. For every study I will obtain the necessary ethics approval.

4 CHALLENGES

As part of my doctoral research, I face interesting challenges con-cerning the conceptual, technical, and methodological aspects ofmy work. These challenges are derived from the qualities of theWeb as my research context and exploratory information seeking asthe targeted task. I plan to approach and address these challengesthrough iterative technological innovations and case studies.

4.1 FlexibilityThe first major challenge for my research is primarily based on thediversity and dynamics of the Web as an information space requir-ing more flexible visualization and interaction techniques.

Data dynamics. Considering, for example, collaboratively editedreference pages on Wikipedia and streams of status updates onTwitter, web-based information spaces can be highly dynamic. Typ-ically, either existing entries are changing or new items arriving.Visualizing such rapidly changing information spaces and datasetsrequires visual representations that adequately indicate change andhelp information seekers make sense of it.

Resource diversity. Web resources can have multiple dimen-sions, e. g., location, rating, tags, that can be harnessed for visu-alization and information seeking. While the set of dimensions

2

Presented during the VisWeek 2010 Doctoral Colloquium.

2009 2010 2011

Literature Review

Research Internship

Web Study

Candidacy

Case Study 1 Case Study 3

Initial Exploration Case Studies Framework

Thesis

Case Study 2

2012

Courses

Research Internship

Figure 2: My doctoral research can be divided into three main phases: initial exploration, case studies, and framework.

typically is constant for a given web community, across the wholeWeb, there is a large diversity of entities that is associated withan equally diverse selection of dimensions. For example, photos,status updates, and reviews have only very few dimensions in com-mon. Designing visual ways for exploring the whole Web in itsdiversity requires more flexible ways of interacting with resourcesand visualizing them.

4.2 Scalability

Another challenge for my doctoral research is based on the extentof the Web as an information and software platform accessed bymore and more people with an increasingly diverse set of devices.

Extent of Web. The number of resources on the Web has reachedthe point where indexing and making it accessible in simple rankedlists has become a great challenge by itself. More sophisticatedvisual exploration systems require even more computational powerand strategies to provide responsive interfaces.

Platform heterogeneity. The Web has become an increasinglycomplex, yet powerful software platform. Developing web applica-tions involves a broad range of programming languages, standards,and machines running as web browsers, servers, and third-party ser-vices. With the emergence of powerful, web-enabled smartphonesand large interactive surfaces the spectrum of web clients is widen-ing far beyond laptops and desktop computers. The heterogeneityof the web ecology requires interfaces to scale between differentscreen resolutions and account for diverse interaction affordancesallowing click, touch, and other types of input.

4.3 Validity

In order to understand how exploratory information seekingchanges with interactive visualization, I need to develop and refinean adequate methodology that leads to valid results.

Factors. The study of information seeking and visualization in-volves a wide range of possible factors on both sides, system andhuman. Isolating the effects of certain factors can be challengingas information seeking can be studied at many levels and granulari-ties. The challenge is to design studies and choose the most relevantfactors to answer my research questions.

Insight. I am particularly interested in studying how interac-tive visualization can affect the search experience of informationseekers. The insight that I aim to get with my research focuseson the role of visualization for more exploratory types of informa-tion seeking. Exploration could be described in terms of curiosity,orientation, and interest. The challenge is to study these arguablyill-defined concepts in the context of visual exploration.

5 TIMELINE

My PhD research is divided into three phases: initial exploration,case studies, and framework (see Figure 2).

5.1 Initial Exploration (12 months)During the first phase of my doctoral research I built the founda-tion for my research by considering existing information seekingapproaches and investigating the potential of visualization for ex-ploring information. As part of this phase, I have undertaken aweb-based study to observe usage and performance of an earliervisual search prototype I developed [9] (see Figure 3). The dataanalysis of logs and survey submissions hint at the potential of vi-sual exploration and suggest exciting ideas for addressing interfaceflexibility and system scalability.

Figure 3: During a web-based study over 750 participants used thisvisual exploration interface, about 200 people submitted feedback.

Throughout my research, I will continuously review literature oninformation seeking and visualization in order to build on ongoingadvances. Particularly during this first phase I am looking at ap-proaches where visualization is used to navigate complex informa-tion spaces having led into my first case study and the formulationof my research proposal.

5.2 Case Studies (24 months)The main part of my PhD work are three case studies of which thefirst one on visualizing microblogs is partially completed with a fullpaper conditionally accepted at this year’s InfoVis conference and auser study in preparation. I started this case study during an intern-ship at IBM Research where I explored the use of web-based in-formation visualization for visually summarizing digital backchan-nels during large-scale events (see Figure 4). This work sheds lighton the idea of evolving visual and topical views for continuouslychanging live data and addresses both challenges in terms of visualrepresentation, interaction techniques, and web performance.

The other two case studies are currently being planned with thegoal to address my research questions and challenges. It is not de-cided at the moment which datasets and domains would be mostsuitable to study the idea of visual exploration for web-based in-formation seeking. My plan is to undertake a second research in-

3

Presented during the VisWeek 2010 Doctoral Colloquium.

Figure 4: The Visual Backchannel interface visualizes live Twitterposts during large-scale events along multiple interactive facets.

ternship and undertake one of the case studies possibly in collabo-ration with an external institution involved in an example domainthat could be interesting for visual exploration.

5.3 Framework (6 months)During the final phase of my PhD research I will summarize de-sign experiences and research findings from the initial explorationand the case studies into a conceptual and pragmatic framework ofvisual information exploration that will become part of my thesis.

6 CONCLUSION

With this research I intend to broaden the understanding of visual-ization for information seeking and how visual exploration relatesto existing interaction techniques such as browsing and searching. Iwill undertake several case studies to establish practical guidelinesfor designing and implementing web-based interfaces for visual in-formation exploration. The participation in the doctoral colloquiumwould be a great opportunity for me to get feedback on my researchplan and the selection of my next case studies.

REFERENCES

[1] C. Ahlberg and B. Shneiderman. Visual information seeking: Tightcoupling of dynamic query filters with starfield displays. In CHI ’94:Proceedings of the SIGCHI Conference on Human Factors in Com-puting Systems, pages 313–317. ACM, 1994.

[2] R. Baeza-Yates and B. Ribeiro-Neto. Modern Information Retrieval.Addison Wesley, May 1999.

[3] M. J. Bates. The design of browsing and berrypicking techniques forthe online search interface. Online Information Review, 13(5):407–424, 1989.

[4] M. Bostock and J. Heer. Protovis: A graphical toolkit for visualiza-tion. TVCG: Transactions on Visualization and Computer Graphics,15(6):1121–1128, Nov/Dec 2009.

[5] A. Broder, E. F. Churchill, M. Hearst, B. Pell, P. Raghavan, andA. Tomkins. Search is dead!: long live search (panel). In WWW 2010:World Wide Web Conference, pages 1337–1338. ACM, 2010.

[6] S. K. Card, J. D. Mackinlay, and B. Shneiderman. Using vision tothink, chapter 1: Information Visualization, pages 1–34. MorganKaufmann, 1999.

[7] S. K. Card, G. G. Robertson, and W. York. The WebBook and theWeb Forager: an information workspace for the world-wide web. InCHI ’96: Proceedings of the SIGCHI Conference on Human Factorsin Computing Systems. ACM, 1996.

[8] E. C. Clarkson, K. Desai, and J. D. Foley. ResultMaps: Visualiza-tion for search interfaces. TVCG: Transactions on Visualization andComputer Graphics, 15(6):1057–1064, Nov/Dec 2009.

[9] M. Dork, S. Carpendale, C. Collins, and C. Williamson. VisGets:Coordinated visualizations for web-based information exploration and

discovery. IEEE Transactions on Visualization and Computer Graph-ics, 14(6):1205–1212, Nov/Dec 2008.

[10] J. English, M. Hearst, R. Sinha, K. Swearingen, and K.-P. Yee. Flexi-ble search and navigation using faceted metadata. In SIGIR ’02: con-ference on Information Retrieval, pages 11–15, 2002.

[11] J. Harris and S. Kamvar. We Feel Fine. http://www.wefeelfine.org/(Retrieved 2010-03-21), 2006.

[12] M. A. Hearst. TileBars: Visualization of term distribution informa-tion in full text information access. In CHI ’95: Proceedings of theSIGCHI Conference on Human Factors in Computing Systems, pages59–66. ACM, 1995.

[13] S. Jones. VQuery: A graphical user interface for boolean query spec-ification and dynamic result preview. In UIST 1999: Symposiumon User Interface Software and Technology, pages 143–151. ACM,November 1999.

[14] C. C. Kuhlthau. Inside the search process: Information seeking fromthe user’s perspective. Journal of the American Society for Informa-tion Science, 42(5):361–371, 1991.

[15] B. Lee, G. Smith, G. G. Robertson, M. Czerwinski, and D. S. Tan.FacetLens: exposing trends and relationships to support sensemakingwithin faceted datasets. In CHI ’09: Proceedings of the SIGCHI Con-ference on Human Factors in Computing Systems, pages 1293–1302.ACM, 2009.

[16] J. D. Mackinlay, G. G. Robertson, and S. K. Card. The perspectivewall: detail and context smoothly integrated. In CHI ’91: Proceedingsof the SIGCHI Conference on Human Factors in Computing Systems,volume 27, pages 173–176, 1991.

[17] P. Pirolli and S. K. Card. Information foraging in information accessenvironments. In CHI ’95: Proceedings of the SIGCHI Conference onHuman Factors in Computing Systems, pages 51–58. ACM, 1995.

[18] Z. Pousman, J. T. Stasko, and M. Mateas. Casual information visu-alization: Depictions of data in everyday life. TVCG: Transactionson Visualization and Computer Graphics, 13(6):1145–1152, Nov/Dec2007.

[19] A. Rivadeneira, D. M. Gruen, M. J. Muller, and D. R. Millen. Gettingour head in the clouds: Toward evaluation studies of tagclouds. InCHI ’07: Proceedings of the SIGCHI Conference on Human Factorsin Computing Systems, pages 995–998. ACM, 2007.

[20] G. G. Robertson, J. D. Mackinlay, and S. K. Card. Cone trees: Ani-mated 3d visualizations of hierarchical information. In CHI ’91: Pro-ceedings of the SIGCHI Conference on Human Factors in ComputingSystems, pages 189–194. ACM, 1991.

[21] D. Rosenberg. Early modern information overload. Journal of theHistory of Ideas, 64(1):1–9, 2003.

[22] A. Spoerri. InfoCrystal: A visual tool for information retrieval. InVIS ’93: Conference on Visualization, pages 150–157. IEEE Com-puter Society, 1993.

[23] M. Stefaner, T. Urban, and M. Seefelder. Elastic lists for facet brows-ing and resource analysis in the enterprise. In DEXA 2008: conferenceon Database and Expert Systems Application, pages 397–401. IEEEComputer Society, 2008.

[24] F. B. Viegas and M. Wattenberg. Artistic data visualization: Beyondvisual analytics. In Online Communities and Social Computing, vol-ume 4564 of Lecture Notes in Computer Science, pages 182–191.Springer, 2007.

[25] M. Wattenberg. Baby names, visualization, and social data analysis.In InfoVis 2005: Symposium on Information Visualization, pages 1–7.IEEE Computer Society, 2005.

[26] R. W. White, B. Kules, S. M. Drucker, and m.c. schraefel. Support-ing exploratory search: Introduction. Communications of the ACM,49(4):36–39, 2006.

[27] M. L. Wilson and m.c. schraefel. Improving exploratory search inter-faces: Adding value or information overload? In HCIR 08: Workshopon Human-Computer Interaction and Information Retrieval, pages81–84, August 2008.

[28] J. A. Wise, J. J. Thomas, K. Pennock, D. Lantrip, M. Pottier, A. Schur,and V. Crow. Visualizing the non-visual: Spatial analysis and inter-action with information from text documents. In InfoVis 1995: Sym-posium on Information Visualization, pages 51–58. IEEE ComputerSociety, 1995.

4