Week 8: Readings and Comments

by Zach Tomaszewski

for LIS 699-2, Fall 2001, supervised by Dr. Luz Quiroga

This was a somewhat flailing, energy-less week.

First off, I caught up on some of the ASIS SIG-IA digest postings. It was entertaining because most of the big names--Lou Rosenfeld, Jared Spool, Steve Krug--were posting. This field is still small enough to feel intimate. Some of the topics this week were:

• How best to label download speeds -- modem/T1, or 50k/100k, or slow/fast? What makes the most sense for new or basic users?
• There was some discussion of how faceted classification is different from fields in a relational database. There is a lot of similarities, but the general thread of it all was that faceted classification is broader than just a database implementation. But the similarities are interesting for implementation purposes. This also led to mention of using faceted classification as a way to provide different navigational or task interfaces depending on users or needs, which is similar to what my project is working on.
• There were some ponderings on whether the web has blurred the distinction between documents and applications, between data and tasks. This seems to go back to the discussion a couple weeks ago about GO and ABOUT use of search engine--as a navigation aid and an information retrieval tool. With interactivity, forms, and navigation, there do seem to be some grey areas. My personal feeling is that the two are still largely distinct (or should be), but, as with all categories, there are those fuzzy borderlands.
• Starting with a long thread on Amazon's interface, the topic of "tabs" came up. Nielsen's quote view is that tabs should only be used to change the view or focus of an information space, not as navigation divisions. A few people pointed out that users probably don't give it all that much thought. Considering how much testing Amazons purportedly does, their choices are probably well backed with empirical user data.
• Another interesting point was whether less navigation is a good thing. In other words, should information be squeezed onto fewer pages and have webpages be more like interfaces?

Another thing that I gave some thought to this week was Fred Riggs's proposal. I missed the last part of our last meeting, so perhaps some of my questions were answered there. Most of my concerns involve the social aspect of users generating content, which is something that relates to my current project. With the lexisaurus, it sounds like he hopes to pass the idea off to a larger company with the component parts already in place, such as the thesaurus and the dictionary. Otherwise, if he was planning to implement it himself, I'd wonder where he'd get the thesaurus and the definitions.

This problem is more evident with the ETL. I think the greatest use of this tool will be when numerous lists have been combined into a list that covers multiple disciplines. It is between disciplines that I see the most conflict and ambiguity between term meanings, and so the place where this tool would be most helpful. However, before this "uber-list" can be compiled, each field must create and publish their own list. This is where I see problems. First of all, which organization will assume responsibility for a discipline's list? What if more than one organization wants to assume the role? Will different list organizations use the same procedures to generate lists? How will they resolve disputes on term meanings? The structure of the different organizations could certainly effect the quality or granularity of the lists they produce.

Secondly, I find it odd that most people would want to generate their own personal term list. They already know what their own terms mean for them. The only time I see people wanting to export lists is when they already know of a term conflict and so want their meanings to be evident to (and possibly imposed upon) the rest of the world.

In short, I find the list generation to be the biggest problem. Also, interfacing and pulling records from a variety of online sources could be problematic on the implementation level. Possibly a large part of the query process could be run from a central server where updates could be made whenever constituent sites change their structures. "For site maintenance ye shall always have with you."

I also briefly read on faceted classification. I was surprised to learn how much relationships were stated between facets. There were not simply "tags" for different attributes. I found similarities between those systems covered and more traditional classification schemes. This is probably because most of these systems had to arrange works in a single place in the system, even if that system was faceted. With computer indices, freely faceted systems are possible.

I also worked on Wayfinding in Architecture, but have no real comments this week.

Post-discussion Issues:

• Discussed designing for more permanent URLs verses using topically-based URLs that mirror the site's structure. While the latter may be more helpful for navigation, the former causes dead links across the web when the site is restructured, negatively impacting traffic.
• Again we pondered how to get a grip on growing, dynamic web content. Is this really possible without frequently redesigning the site as new content comes available?
• Luz mentioned that she would in fact have a use for a personal ETL. Cool; then perhaps there is more hope for Riggs's project than I first thought.

To Week 9 →