Visualizing Knowledge Domains: ASIS&T 2002

Helen Barsky Atkins from HighWire showed features from their search interface. It’s a large archive of free life science articles, plus indexing for many that require a subscription to view the full text.

The features she showed let users browse related information via citation linking and related subjects in what they called a “topic map.” The topics are generated from a “peer reviewed taxonomy.” (Hmmm, wonder how that works.) Entrieva (Semio) figures somewhere in here.

This “topic map” was one of those hyperbolic trees that I’m very skeptical about. My view might be skewed from years of java implementations to generate hyperbolic trees that crashed my poor mac pre OS X, but I think we should just agree that they are useless and move on. She said that it had a huge “wow” factor and that was about it.
The citation map displays all articles either citing a particular article in a list of hits, or cited by the article. I should start collecting examples of ways to display this sort of information. Citeseer just displays it with plain text in logical groupings. Highwire has little colored ovals with arrows pointing to your selected article. I’m not sure it really adds that much to it to display it this way. They do add a color scheme that indicates which of these articles are frequently cited by making their ovals yellow so you can see at a glance which of the papers have had the greatest impact.

I think they just need to remove a lot of unnecessary graphics from their display so the more informative visuals don’t have so much competition. For example, for each citation, the cover of the journal it appears in is depicted in addition to the title of the journal appearing twice. There is a more condensed view you can select (if you notice that option among all the visual noise). In the condensed view, each citation has a little arrow graphic where the alt text is the extended citation. I hate that particular abuse of the alt tag. It only seems to be proliferating, too.

She also showed a feature that I can’t seem to find on the live site. Apparently the terms assigned to each article are weighted. Once you select an article, you can use these assigned weights to refine your search. The interface uses sliders which appear next to each of terms assigned to a selected article. I think it’s an extension of the “more like this” feature most search engines have, but it’s actually “more like this, but a bit more in favor of topic X.” I like that. I kind of have a thing for sliders, though. It could be a useful feature, or I may have been seduced by a sexy widget.