|
|||
|
|
|||
|
Will there come a time when we can view numbers and text as shapes and colours and understand information space as we do physical space? Some years ago (April 1996), James Fallows wrote an article in the Atlantic Monthly titled Computers: Navigating the Galaxies in which he imagined “surfing the contours of information space" "As you look at your computer screen, you seem to be flying at low altitude above a flat midwestern plain marked with billboards announcing different topics - sports, news ... If you descend to get a closer look, you can see the names of subtopics behind each billboard, and then the articles you're looking for." This hasn’t happened yet, but it will. Information visualization – seeing information represented in pictures or shapes – “increases our bandwidth of access”. We comprehend more information faster. James Wise of Integral Visuals sees information visualization as the emerging science. Wise spoke at the Infonortics conference, Search Engines and Beyond in April 1999, about a naturalistic approach to visualization. Imagine documents as stars in the sky, consider sedimentary layers, look for topographies – all of these have the naturalistic view. Two applications that stand out as candidates for information visualization are the subject directory – displaying the hierarchy of topics, and document text analysis – highlighting the clusters of documents on a topic and their relevance to the query. Some trials have faltered. Pointcom, once known for its Top 5% collection of web sites, tried 3D representation of its subject directory. The software was early VRML and very slow but you could pick out the big blocks – Business, Arts – and get into those spaces. AltaVista used to have a refine function by which you could get a graphical view of the core terms that occurred in a document set and see the strength of relationships between them. Alas, Altavista discontinued Refine in April 1999 and offered no explanation. Today, visualization aids are more likely to be incorporated into a
product for the Intranet than a service on the Internet. Fortunately, we can at
least get an idea of the new tools from demos on the Web. The following are examples of
products that you can view today. All but one are commercially available and have demos.
To run them, however, may require that you have version 4.0 or above of Communicator or
IE, with Java enabled and that your firewall does not block Java applets. At least
56 kbps would also be good. Products Using Information Visualization Semio Corporation: Building Web and Intranet Directories
Semio Builder software analyzes bodies of text (referred to as corpora) to identify the main concepts. It displays these in a concept map where each concept is a node with links to other nodes; there are sub-nodes at deeper levels. A demo of SemioMap Discovery is available at the Web site with several sample selections from the catalog. I recommend the Starr Report because of the complex relationships it illustrates – data ones, that is. Semio also has a taxonomy product which will create browse-able directories from the analysis.
Themescape is also text visualization software. Using relational topic mapping (RTM) technology, Themescape reads large collections of documents and organizes the content by topic as a topographical map. There are hills and valleys and contours. Peaks indicate a concentration of documents. Other visuals in the landscape are labels, points, flags (some terrains will look like an 18 hole golf course); to navigate, you zoom in and out. The demos at the site are excellent. NewsMaps, (http://www.newsmaps.com), uses ThemeScape to
present the news. Each day NewsMaps analyzes news stories and displays the main topics as
maps with links to the topics and the actual stories. Newsmaps likens the maps to a visual
Table of Contents. The Web site has several you can use today covering the Kosovo
Crisis, US and Global news, Business news, and Technology news. The collection
is current, searchable and free. This is easily the most impressive product to come to the
Web in some time. Inxight Inxight (pronounced insight) is a Xerox New Enterprise business initiative. It aims to “make information make sense” and it certainly accomplishes this through its natural language processing and visual aids. Most notable is the use of the hyperbolic tree to navigate subject directories or any collection of related files such as a web site. There are several fascinating applications of the hyperbolic tree in demo form at their web site. (Or go directly ). See the Louvre, the Library of Congress, Yahoo, Netcenter, the British Monarchy – and many more. The Wedge from Australia National University With the Wedge you walk into the theatre and the information swarms around you in 3D images. Mind, you need special glasses to get the stereoscopic vision. This is a research project at ANU Supercomputer Facility. It may be a while before it comes to a theatre near you. Dr Link from Manning and Napier Information Services
DR-Link (Document Retreival using Linguistic Knowledge) has an
extensive online collection of journals in full-text. Its distinguishing features are full
natural language processing and the use of graphs and charts. Search results to a
text query can be viewed as graphs where the authors, subjects, and organizations named in
the document set are ranked. Another option shows frequency of documents by date as a
graph. Not only is it much easier to determine where to look first, but the graph imparts
more information about the topic - when it was most in the news, who was most
mentioned in connection with that topic. Daily Diffs from InGenius DailyDiffs records the changes at Web sites. The display shows the parts of the page where text has changed over the period, day by day. DailyDiffs monitors about 4000 pages at its Web site. These are grouped by topic. To see the history, work your way down the subject tree until you reach the DailyDiffs page listing the pages they are tracking. Overall the display is quite nicely done but it is the Short History that really impresses with its pictorial representation of page changes. Conclusion James Wise reported that 65 universities have research projects
related to document and text visualization involving departments for computer science,
library science, humanities, psychology, and geography. One day that research will surge
to market in new commercial applications. Watch for it. References:
Sessions with comment about information visualization:
Copyright © 1999 SLA. All rights reserved. |