of Order in a Chaotic World: Using Verity's Intelligent Classifier to Create a
Corporate Taxonomy for an Intranet Search Engine
5:40 - 6:00
Levels of Order in a Chaotic World: Using Verity's Intelligent
Classifier to Create a Corporate Taxonomy for an Intranet Search Engine
M. Proserpio, Technical Library & Strategic Web Information Framework &
Technology, Sandia National Laboratories
Sandia National Laboratories' inception
of the Verity intranet search engine has not met users expectations. Users have
complained the search engine is not an effective tool in searching the Labs approximately
150,000 plus web pages. The search results, based on full-text indexing, produce
too many hits. In addition, these results are often too broad for users to sift
through, understand or use. One of the issues contributing to the poor results
has been the inconsistent application of standards (e.g., metadata and information
structure) to the extensive number of pages on the corporate intranet. We believe
categorized results, based on a corporate information taxonomy, coupled with improved
application of standards will yield a marked improvement for users seeking information
on the Sandia National Laboratories corporate intranet.
In an attempt to
enhance the search engine's search results and its use, the Technical Library's
staff developed and implemented a subject-based categorization for the search
engine using Verity's Intelligent Classifier software. Over a 6 month period,
a team of four information professionals and one programmer analyzed and categorized
external and internal web collections (approximately 150K web pages), developing
a corporate information taxonomy. There were numerous challenges faced by the
team, including: different styles in categorizing highly technical content, administrative
deadlines, underestimating the magnitude of the project, and addressing information
issues out of the scope of the project (e.g. limitations of the Verity spider),
etc. This new corporate taxonomy is due to rollout in January 2003.
* Each team member's style differed based on their understanding and analysis
of the content.
* Establishing a taxonomy in a corporate culture that does
not emphasis standards for metadata or web-posted information.
new tools, from the ground up. Migrating to new versions of the tool. Limitations
of the tools.
* Weaving existing small-scale taxonomies and subject-categorizations
of corporate information into the new, large-scale taxonomy.
* Deciding how
far down to develop a taxonomy that is built on a dynamic content base.
Promoting the work of the information architects on this project.
* How to
improve the information culture of the corporation in a positive way.
(To be announced. In order to conduct relevant usability analyses, we intend to
conduct focus groups, track metrics, and solicit feedback from users on a regular
View the Conference Schedule