Week 7: Readings and Comments

by Zach Tomaszewski

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

Between various other engagements, I got very little reading done this week. I'm still working on Wayfinding in Architecture and Principles of Animal Taxonomy.

One of the things I did this week was update my LIS site a little. In doing so, I realized that some of our discussions about the difficulty of organizing online learning courses might be boiled down to the fact that the structure needs to come before (or at least with) the content. One of the exercises suggested by Rosenfeld is to take all your content, modeled on notecards, and sort it into piles. These content piles can be used in determining sturcture and labels. However, there are many systems where the content is created by a variety of users and goes through no real standardizing point. Also, the structure is created on-the-fly; the structure has to be predicted before the content exists.

For instance, my web site began with two or three pages, and an index page. There is no real system. Some classes have their own subdirectories if is it seems there will be a large number of assignments for that class. Files names, and so URLs, are very haphazard. Sometimes I wish they followed a specific format based on the class and assignment number; sometimes I'm glad the names are related to the content. And often either the class number or the assignment does not fit well into such a file-naming system. Also, there is no standardized style across the site, though there is a good attempt at this. These issues are relatively minor, but my site is still relatively small--maybe 40 to 50 pages. I think the seeds of disaster have been planted. But how do we provide a structure before we can see all the content we need to organize? Ah, there's the rub!

Related to this issue, I started thinking about how to provide a web interface to my current ICS691 project. (The current draft can be seen here.) This is another case where the content will largely be created by different users of the system.

Basically, the system will contain multimedia records for certain animals--polar bears, sea otters, emperor penguins, etc. Users will be able to use a variety of different kinds of classification systems to access these different animal records. Some privledged users (Creators) will also be able to add their own classifications to the system.

Already, we have two user groups. Creators will have to seach and use the system in every way that regular users will, but will also have certain functions and task-specific needs related to creating and adding classifications.

Searching will probably be the primary mode of finding classifications or (directly) animal records. However, what should be indexed for searches in these cases? Should nodes and animal records be accessible without traversing any specific classification root? And how will browsing be implemented?

The biggest problem I forsee is consistent labeling. I'd like to avoid duplicate classifications masking under slightly different names. What kind of classification of classifications could be done, before knowing what users will create? Possibly "Classifications based on intrinsic animal characteristics," "Classifications based on animal's habitat," "Classifications based on an animals relationship to humans," etc. Already I can imagine overlaps and gaps in such a system. Yet if the system is to be of any size, some sort of topical division of classifications will be needed.

This week, I'd like to discuss ways to design a structure for a multi-user, multi-author system. Any insights would surely be applicable to such an implementation project.

Post-discussion Issues:

To Week 8 →