NAHUM GERSHON, STEPHEN G. EICK AND STUART CARD design

RVI/Rutt Video Interactive Attract Loop Sequence Information

Introduction

Visualization provides an interface between two powerful information processing sys-

tems—the human mind and the modern computer. Visualization is the process of

transforming data, information, and knowledge into visual form making use of

humans’ natural visual capabilities. With effective visual interfaces we can interact with

large volumes of data rapidly and effectively to discover hidden characteristics, pat-

terns, and trends. In our increasingly information-rich society, research and develop-

ment in visualization has fundamentally changed the way we present and understand Vinteractions...march + april 1998 9

large complex data sets. The widespread and areas [10–12, 14]. Information visualization fundamental impact of visualization has led to [6, 14] combines aspects of scientific visual- new insights and more efficient decision ization, human–computer interfaces, data making. mining, , and graphics. In contrast to Much of the previous research in visualiza- most scientific data [9], information visualiza- tion was driven by the scientific community tion focuses on information, which is often in its efforts to cope with the huge volumes of abstract. In many cases information is not scientific data being collected by scientific automatically mapped to the physical world instruments or generated by enormous super- (e.g., geographical space). This fundamental computer simulations [9]. Recently a new difference means that many interesting classes trend has emerged: The explosive growth of of information have no natural and obvious the Internet, the overall computerization of physical representation. A key research the business and defense sectors, and the problem is to discover new visual metaphors deployment of data warehouses have created a for representing information and to under- widespread need and an emerging apprecia- stand what analytical tasks they support. tion that visualization techniques are an essen- Information can come in huge quantities tial tool for the broad business and technical and in fast streams, creating an information communities. avalanche. The largest information space is Information visualization deals with new perhaps the World Wide Web (WWW), classes of data and their associated analytical which contains millions of pages. Information tasks in business and information technology visualization needs to enable users (e.g., in the commercial and the defense sectors) to get information fast, make sense out of it, and reach decisions in a rela- tively short time.

Developments Document and Large bodies of text require the users to read many docu- ments, understand them, and make sense of the infor- mation in them. Represent- ing the information in a visual form could allow the user to browse through this ocean of information and find interesting pieces of text. Documents have con- tents and histories. In many cases both could be repre- sented by visualization tech- niques. Looking at relationships within the con- tents of large sets of docu- ment histories is difficult to trace. A visualization tool, Figure 1 SeeSoft S.G. Eick, J.L Steffen, and E.E Sumner, Lucent Technologies’ Bell Lab themescapes, was developed

10 interactions...march + april 1998

at Pacific Northwest National Lab [16]. A cluttered and become visually confusing themescape is an abstract three-dimensional for larger sets. One promising approach for landscape of information extracted from the increasing the density of information on relationships among documents. Eick and graphs involves using distortion lenses, Nahum Gershon others at Lucent Technologies/Bell Labs devel- enabling the viewer to see the detail and The MITRE oped a method to represent properties of soft- the general context. A survey of the field as Corporation, ware programs comprising millions of lines well as a new contribution that extends dis- McLean, VA [4]. This system, SeeSoft, uses an arrangement tortion techniques from two to three [email protected] of colored columns and rows (see Figure 1) to dimensions is given by Carpendale et al. represent the frequency of use or other attrib- [3]. Nevertheless, studies elucidating when Stephen G. Eick utes of the particular lines of source code. This distortion techniques are effective are Bell Labs/Lucent system is currently being used to help detect needed. Technologies lines of code associated with the Year 2000 Naperville, IL problem [5]. Hierarchies eick@research. The traditional way of depicting hierar- bell-labs.com Distortion Techniques chical information is to structure it in a Many types of information involve relation- tree-like node and link . For large Stuart Card ships. One common way to visualize struc- trees, however, these diagrams rapidly Xerox Palo Alto tured relationships is to use a graph, with become cluttered and unusable. One of the Research Center nodes representing the entities and links the earliest instances of information visualiza- Palo Alto, CA relationships between the entities. Graphs tion—the cone tree, developed by Card et card@.xerox.com work well for small information sets (tens to al. [14]—is a 3D representation of hierar- hundreds of nodes and links) but are easily chical information (see Figure 2). Tree

Figure 2 Cone Trees G.G. Robertson, S.K. Card and J.D. Mackinlay (Xerox PARC)

interactions...march + april 1998 11

Figure 3 The user performs a recomposition of separate pieces of data located in separate collections. Newly recom- posed collections of shipments are dragged to the original routes depicted on a , where they are shown at their destinations [8]. J. Kolojejchick, S.F. Roth, and P. Lucas (CMU and MAYA Design Group, Inc)

[8] have described new ways to create a coordinated suite of basic tools and specialized information appliances. One important contribution of Kolojejchick et al. is to combine many basic analysis and reporting tools into an integrated information workspace (see Figure 3).

World Wide Web The World Wide Web has enabled the visualization community to pro- vide visualizations to a large-scale audience. It is already changing the way visualizations are delivered to users. Information visualization can also facilitate the user interaction with information residing on the WWW. A good example is WebBook, by Card et al. [2]. WebBook allows rapid interaction with objects at a level higher than pages (e.g., by placing a series of , developed by Johnson and pages in a booklike arrangement). A second Shneiderman, visualize hierarchical informa- example is Web Forager, an application that tion on a two-dimensional display by filling in embeds WebBook and other objects in a hier- available screen space [7]. archical workspace (see Figure 4). Using the Web to deliver visualizations Integrating Information Across Multiple means that visualizations are no longer canned Applications products predetermined by their provider. Over the years, researchers have developed a Data could be sent over the network and the diverse range of visualization techniques suit- visualization can be performed on the user able for specific data and information types. (client) side. Not only can this save time in Real-life situations may require visualization of importing the visualization product over the information and data from diverse sources, network, it can also enable the user to tailor whereas the application might require the use the displayed visualization to suit his or her of different visualization tools. The lack of a problem, needs, or capabilities. These systems common set of operations and ways to inte- can be implemented using languages and stan- grate information across multiple applications dards such as Java and Virtual Reality is a long-standing problem in research and Modeling Language (VRML). Another advan- application development. Kolojejchick et al. tage of Web representation is the use of hyper-

12 interactions...march + april 1998

links, visually interconnecting points in the bases using data mining to provide deci- visual representation to other objects. Some of sion support in commercial and other these issues, together with some examples of environments [1]. new and potential applications, were addressed by Rohrer and Swing [15]. Challenges Understanding the Advantages and Applications to Real-World Problems Disadvantages of the New Media The importance of information visualization The media of visual computing and display are is its potential to solve real-world problems far quite new, and we do not understand well their beyond a purely academic interest. It is advantages and disadvantages. Many developers exciting to note that this relatively new disci- and users currently relate to the new medium of pline has already penetrated the commercial visual computing and display as if it were a market. Some examples of commercial appli- replica of paper, to which we have grown accus- cation include tomed over the past thousand years. These new • Solving problems in the financial market technologies, however, allow us to do certain [17], things beyond what is possible with paper (see • Using applications of ideas developed at e.g., [6]). It will take us time to develop suffi- Xerox PARC (e.g., Wall, Cone cient understanding of these new media. Tree, Wide Widgets) to forge the next generation of user interface that will be Emerging Discipline more visual than current ones [13], and As a discipline, information visualization is still • Using information visualization to repre- emerging, tracking the revolution in net- sent knowledge extracted from large data- working and computing. It is being practiced

Figure 3 Web Forager S.K. Card, G.G. Robertson, and W. York (Xerox PARC)

interactions...march + april 1998 13

by skilled practitioners who are hand crafting 3D graphics and animations. The challenge, the current systems. An emerging discipline however, is how best to exploit this forth- progresses through four stages. First, it starts as coming capability. Currently, we do not always a craft, practiced by artisans using heuristics. understand when 3D is more effective and is Second, researchers formulate scientific princi- not more effective than 2D. As better software ples and theories to gain insights about the is created to make producing visualizations processes. Third, engineers eventually refine easier, it will be important to use these new these principles and insights into production capabilities when they are appropriate and rules. Fourth, the technology becomes widely information-conveying. available. For information visualization many of these stages are happening in parallel. Human- and Usage-Centered Visualization Ease of Use Tailoring visualization systems that will be In contrast to , which seriously based on human capabilities of per- focuses on highly trained scientists, interfaces ception and information processing is another created for manipulating information may be challenge. We need to better understand how broadly deployed among a diverse and poten- human beings interact with information, how tially nontechnical community. The demand they perceive it visually and nonvisually, how for good and effective visualization of infor- the mind works when searching for both mation embraces all walks of life and interests. known and unknown information, and how it This user community is diverse, based on solves problems. A good human-computer varying levels of education, backgrounds, interaction is a must, but it is not enough. We capabilities, and needs. We need to enable this also need to better use the information we diverse group to use visual representations have about how humans understand and that will be tailored to their specific needs and interact with information and the perceptual the specific problem at hand. system in designing visualization systems.

PERMISSION TO COPY WITHOUT FEE, Another human-based challenge is the need to

ALL OR PART OF THIS MATERIAL IS Application of Visualization to Real- learn how to create flexible user interfaces,

GRANTED PROVIDED THAT THE COPIES World Problems navigation tools, and search methods that will

ARE NOT MADE OR DISTRIBUTED FOR How to use information visualization to solve be appropriate for each of the existing types of

COMMERCIAL ADVANTAGE, THE ACM real-world problems is one of the major chal- users, applications, and tasks.

COPYRIGHT NOTICE AND THE TITLE lenges of information visualization technology.

OF THE PUBLICATION AND ITS DATE The motive of a large part of the research and Text vs. Images

APPEAR, AND NOTICE IS GIVEN THAT development in this area is academic interest Recent developments in visual display com-

COPYING IS BY PERMISSION OF THE rather than applicability, thus creating a “bag of puter hardware on the one hand and com-

ASSOCIATION FOR COMPUTING tricks” (Ken Boff, private communication). An puter graphics and visualization methods and

MACHINERY. TO COPY OTHERWISE, alternative and more demanding approach is to software on the other have generated new

OR PUBLISH, REQUIRES A FEE AND/OR study the problem first and then look for interest in images and visual representations.

SPECIFIC PERMISSION. appropriate solutions, for example, using infor- “A picture is worth a thousand words,” goes ©ACM 1072-5520/98/0300 $3.50 mation visualization, if appropriate. We need the popular saying. However, images may to realize that in many real-world applications, have some disadvantages, and words are visualization is just one component of a com- sometimes more effective (or powerful) than plex system rather than a stand-alone entity, pictures. The disadvantages of using images and we need to understand the system and user include difficulty in representing information needs in order to create effective visualizations. clearly; dependency of visual and information perception on memories, experiences, beliefs, Three-dimensional vs. Two-dimensional and culture; and difficulty in making effective With the widespread deployment of three- use of color. To use images effectively in sci- dimensional graphics chips, desktop PCs will ence, education, art, and life in general, we soon be capable of much more sophisticated need to understand the power and frailty of

14 interactions...march + april 1998

images. We need to understand when they are 957–968. equivalent to words, when they are more 5. Fyock, D.E. “Using visualization to maintain large appropriate to represent information than computer systems.” and Applications words, and when they are not. This issue has (July–Aug. 1997), pp. 73–74. become extremely important with the spread 6. Gershon, N. and S.G. Eick. “Visualization’s new of the Web, whose many document authors tack: Making sense of information.” IEEE Spectrum use graphics inappropriately. (Nov. 1995), pp. 38–56. 7. Johnson, B. and Shneiderman, B. “Tree-maps: A The Future space-filling approach to the visualization of hierarchical Research and development in the emerging information structures.” In Proceedings of IEEE field of information visualization has Visualization ‘91, Computer Society Press, Los increased dramatically in the last few years. Alamitos, CA, 1995, pp. 284–291. With hardware and software prices decreasing 8. Kolojejchick, J., Roth, S.F., and Lucas, P. and speed increasing, more diverse groups of “Information appliances and tools: Simplicity and users are going to use this technology. In addi- power trade-offs in the Visage exploration environ- tion, as the use of the Web becomes more ment.” Computer Graphics and Applications (July–Aug. sophisticated and more commonly used, it 1997), pp. 32–41. will have a major impact on the way we 9. McCormick, B.H, Defanti, T.A., and Brown, M.D. deliver information visualization. With the Visualization in scientific computing.” Computer increasing use of visual media, the popula- Graphics 21, 6 (Nov. 1987). tion’s visual literacy will increase, and as a 10. Proceedings of Information Visualization Symposium result users will become more comfortable ‘95. Gershon, N. and Eick, S.G. (eds.), Computer dealing with visuals and thus will get more Society Press, Los Alamitos, CA, 1995. information from them. Developers, in turn, 11. Proceedings of Information Visualization Symposium will include human considerations, creating ‘96. Card, S., Eick, S.G., and Gershon, N. (eds.), more and more usable visualization systems. Computer Society Press, Los Alamitos, CA, 1996. The development of scientific and engi- 12. Proceedings of Information Visualization Symposium neering principles for the generation of visual- ‘97. Dill, J. and Gershon, N. (eds.), Computer Society izations (to users with diverse needs and Press, Los Alamitos, CA, 1997. capabilities) and a methodology for solving 13. Rao, R. “From research to real world with Z-GUI.” problems with information visualization are Computer Graphics and Applications (July–Aug. 1997), badly needed. pp. 71–72. 14. Robertson, G. G., Card, S.K., and Mackinlay, J.D. References “Information visualization using 3D interactive anima- 1. Becker, B.G. “Using MineSet for knowledge dis- tion.” Communications of the ACM 36, 4 (1993), pp. covery.” Computer Graphics and Applications (July-Aug. 57–71. 1997), pp. 75–78. 15. Rohrer, R.M., and Swing, E. “Web-based informa- 2. Card, S.K., Robertson, G.G., and York, W. “The tion visualization.” Computer Graphics and Applications WebBook and the Web Forager: An information work- (July–Aug. 1997), pp. 52–59. space for the World Wide Web.” In Proceedings of CHI 16. Wise, J. A. , Thomas, J.J., Pennock, K., Lantrip, ‘96, ACM Conference on Human Factors in Software. D., Pottier, M., and Schur, A. “Visualizing the non- ACM, New York, 1996, pp. 111-116. visual: and interaction with information 3. Carpendale, M.S.T., Cowperthwaite, D.J., and from text documents.” Proceedings of Information Fracchia, F.D. “Extending distortion viewing from 2D Visualization Symposium ‘95, Gershon, N. and Eick, to 3D” Computer Graphics and Applications (July–Aug. S.G. (eds.), Computer Society Press, Los Alamitos, CA, 1997), pp. 42–51. 1995, pp. 51–58. 4. Eick, S.G., Steffen, J. L., and Sumner, E. E. 17. Wright, W. “Business visualization applications.” “SeeSoft—A tool for visualizing software.” IEEE Computer Graphics and Applications (July–Aug. 1997), Transactions on Software Engineering 18 (1992), pp. pp. 66–70.

interactions...march + april 1998 15