Web-based information visualization

R.M. Rohrer, E. Swing
1997 IEEE Computer Graphics and Applications  
I nformation visualization, an emerging discipline, uses visual means to represent nonspatial, abstract data. To visualize such information, you must map this data into a physical space. Finding the appropriate visual mapping for the task at hand proves vital to producing effective visualizations. Information visualization can often help you find and understand relationships and structure within (seemingly) unstructured data. Recent widespread interest has focused on exploration of information
more » ... isualization techniques and applications for just that reason. At the same time, information has become pervasive thanks to underlying mechanisms such as the World Wide Web (WWW) and corporate intranets. Visualizing Web-based information-either from the WWW or intranets-has become a common application of information visualization. Given these trends, the Web has naturally progressed as a source of information as well as an underlying delivery mechanism for interactive information visualization. To further explore these ideas, developers use tools such as Virtual Reality Modeling Language (VRML), Java, and Web browsers such as Netscape to create Web-based information visualization applications. While a Web-based delivery mechanism offers a number of advantages, it also imposes a number of limitations and problems. A fundamentally new medium for visualization, the Web is changing the way visualization applications are developed, delivered, and used. We have developed a number of Web-based information visualization prototypes and applications by adapting several well-known information visualization ideas and techniques for use within Web environments. Before delving into specific examples, we offer some relevant background about the Web and our use of visualization for analysis. The Web provides a flexible means for linking applications, data, information, and users. To seamlessly interlink associated data and couple visual representations with this data creates opportunity for new approaches to visualization. We use the term Web-based information visualization to describe visualization applications that use the Web as an information source, a delivery mechanism for visualization, or both. In fact, the synergistic effect of integrating the source and delivery aspects of the Web provides strong support for visualization in this new medium. A Web undercarriage affects how you approach visualization. The Web provides a loose coupling between data, users, and applications. While this gives flexibility for remote data access and logical associations between data, it also means limited knowledge transfer between client and server. Visualization applications targeted for the Web need to account for this. Current Web interaction and navigation techniques follow an intuitive point-and-click paradigm that lets users follow associated hyperlinks and drill down to underlying data. Web-based visualization interactions often follow a similar approach and may well evolve into new hyperlink-based interaction paradigms. Web-based visualization lets users customize applications and data representations not originally targeted for each other by dynamically linking Web-based data and visualization applications. Push technologies, which send (push) information to a user's desktop, automatically deliver data and information to users through customized subscription to data services. This flood of information flowing to the desktop may require visualization for succinct presentation and structuring. To be effective, these pushed visualizations will need to integrate disparate pieces of information. Web-based visualization also provides an easy mechanism for incorporating Web-based multimedia data into applications. The Web provides a rich, flexible new medium for information visualization applications. New techniques and paradigms will continue to emerge as we gain experience using the Web. Web is being used to help visualize complex relational information. Here we discuss several Web-based visualization prototypes and applications. Web-based development tools In adapting visualization techniques for use in Web environments, we employ several common development tools, including VRML as a 3D graphics engine and standard Web interfaces such as Netscape for formatted text browsing. Currently, Hypertext Markup Language, or HTML, remains a common standard for providing formatted, interlinked text. Common Gateway Interface, or CGI, provides a method for interfacing Web tools with external gateway programs. These simple Web-based navigation and modeling tools permit using hyperlinks and CGI-triggered database queries to interlink related data and information. In addition, integrating Java with these tools enables the development of interactive, distributed Web applications. VRML has become a popular de facto standard for producing 3D graphics within Web environments. Currently, a simple scripting language defines VRML scenes or object descriptions. Defined as a standard Multipurpose Internet Mail Extensions (MIME) data type, VRML scenes can be accessed through standard hypertext transfer protocol (HTTP) Web server mechanisms. VRML-capable browsers and plug-ins on the client side let you access and interact with VRML content that's published on a Web server. In addition to standard 3D graphics functionality, VRML embeds hyperlinks directly into 3D worlds. Individual objects can be hyperlinked to any Web media type such as text, images, movies, sounds, and other VRML worlds. These embedded hyperlinks can also trigger CGI-based actions such as database queries. The initial version of VRML (VRML 1.0) defines a static scene. Visualizations based on VRML 1.0 prove somewhat limiting because users can only navigate and manipulate scenes as a whole and follow available hyperlinks. However, despite these limitations, a somewhat primitive interaction and dynamic update mechanism can be achieved by using server-side CGI programs for controlling access to data, processing, and broadcasting scene updates to VRML clients. The updated VRML 2.0 standard adds the ability to incorporate dynamic behavior, animation, and user interaction with Java and JavaScript as an underlying computation engine on the client side. This improved version of VRML makes it easier to produce dynamic and interactive Web-based visualizations. Java has become a popular mechanism for providing client-side interaction in Web applications. It provides a more tightly coupled communications and computation mechanism for Web-based applications, which typically have a loose coupling between client and server. Java interpreters-embedded within standard browsers such as Netscape-provide the underlying dynamics engine of VRML 2.0. To date, we have produced our Web-based visualizations by directly generating VRML scenes from an application program. In the future, we plan to adapt our work to higher level toolkits and Java-based solutions. Visualizing Web-based information You can visualize several structures inherent in Webbased information. These structures include Web topology, external structure, and cognitive models. Depending on the application goals and data set characteristics, these Web structures can play a significant role in choosing and designing visualization techniques. The first inherent structure is the topology of the Web. This includes hyperlinks from page to page (media to media) as well as internal structure within documents. Typically, Web authors or content creators define these relationships and links. Traversing these links or topology remains the most common model of Web navigation and browsing. The second inherent structure comes from structure imposed from outside the Web (external structure). External structure is often a logical grouping-for example, an organization's Web page hierarchy grouped and displayed by department or office. This may prove useful to identify the work of various offices as they contribute to the whole organization. Another example of external structure is the physical layout, geographical layout, or communications model of interconnected Web servers. Sometimes this external logical structure closely matches the underlying Web topology. The cognitive model the user perceives comprises the third inherent structure. This structure depicts content or topic relationships that associate two or more pieces of information. These pieces of related information may come from separate topic domains and may or may not be explicitly linked to each other. Users frequently create their own desired cognitive mapping by grouping related information in a personalized structure such as a bookmark, hot list, or Web page. In fact, researchers are continually studying ways to help people track and organize these cognitive relationships. One of the primary strengths of Web-based information systems is that they abstract and hide these various structures from users. For example, you do not need to know the geographic layout of interconnected Web servers to access information on them. However, when producing Web-based visualizations, you can use these inherent Web structures to augment visualization. We often use visualization to understand the WWW and improve Web navigation and search tools. Early work in Web visualization represents the beginning of an important new area of research. Gershon et al. 1 developed methods for visualizing hierarchical Web structures and for visually creating a personal information space linking related views of information. Munzner 2 used VRML-based visualization for displaying the WWW's structure and providing links to underlying data. Andrews 3 developed the Harmony system for navigating and visualizating the Hyper-G Internet information system. Mukherjea 4 used Web-based techniques for visualizing the results of a multimedia Web search engine that finds references to both text-based documents and images. Finally, Alper 5 used a Java-based visualization tool for querying and displaying geospatial data within Web environments. Web-based visualization While visualizing the Web is a common application of visualization, using the Web to deliver visualization has become a more recent trend. Wood and colleagues 6 used Web-based visualization for providing interactive views IEEE Computer Graphics and Applications 53 of environmental data on the WWW. (They also provided a nice taxonomy and reference model for creating visualization within Web environments.) Walton 7 described and referenced a number of Web-based visualization applications that build upon VRML. Web-based visualization has been adapted for numerous, diverse applications such as chemistry, 8 manufacturing data, 9 and analyzing and debugging parallel computing systems. 10 Visualization for analysis Several examples of our work help illustrate the potential benefits and applications of Web-based visualization. Like many institutions, the National Security Agency (NSA) must cope with large amounts of information. We are exploring numerous approaches to understand relationships, trends, and anomalies in large, disparate, multimedia data sources. One of our technology labs-the Community Wide Enterprise Facility (CWEF)-actively explores information visualization techniques to help information analysts understand large, abstract data sets. In addition, like other information-based organizations, NSA is attracted to Web-based technologies. The Web offers a flexible, seamless way to link tools, data, and users. We see our internal Web as a good mechanism for linking such resources. We view visualization as an integral component in data analysis. For very large data sets, visualization may be the only possible approach. Our development efforts help information analysts find important relationships, anomalies, and trends. Visualization techniques will become even more important as the volume of data continues to increase. Visualization may lead directly to analytic discovery or may be used for data reduction. Direct analytic discovery highlights a key knowledge nugget from a large corpus of data. Using visualization as a data reduction tool, analysts can cull away uninteresting portions and then apply more conventional tools to the remaining data. While we do not see visualization as a panacea, we feel that it is an important analytic tool that will become even more crucial in the future. Visualization examples In the following sections, we survey a number of important, well-known information visualization techniques and briefly describe our adaptation and implementation of these approaches to Web-based solutions. This is a snapshot of ongoing work, which we continually mold and evaluate to produce useful Web-based applications. To date, evaluating the effectiveness of these Web-based applications has been subjective. We need more formal usability studies to truly measure their effectiveness within Web environments. However, our initial subjective experience has been favorable. Hierarchical information We often use visualization to understand hierarchical information structures. Examples of information hierarchies include organization structure, computer file systems, interlinked Web hierarchies, and communication hierarchies. Common approaches for visualizing information hierarchy are cone trees, 11 tree maps, 12 and hyperbolic browsers. 13
doi:10.1109/38.595269 fatcat:d2mm6lx5zfeh7bm2ntzbhn2r3e