We all know that it's tax season, but perhaps we should declare it taxonomy season, too? With so much focus on optimizing Web sites for search engine rankings and paid search listings, little attention is devoted to the importance of implementing robust taxonomies to define standardized and consistent metadata for one's content collection. The major Web search engines recognize the need for tagging content found on disparate Web sites in order to improve contextual matching and personalization algorithms, but is a top-down generalized search engine approach going to be able to tackle the task of developing a taxonomy for everything on the Web?

Last week I attended a seminar on taxonomies that was sponsored by Verity, Inc. Taxonomies are a hot topic. Maybe not exciting, but definitely hot. The room was filled with IT and information professionals who are faced with the tricky task of implementing a taxonomy across their institutions' content collections. With the help of a knowledge expert or an information specialist/librarian, creating a robust and dynamic taxonomy doesn't have to be difficult, particularly if tools and prebuilt taxonomies are available. However, all panelists at this seminar agreed that the trickiest aspect of building and implementing an organization-wide taxonomy involves pulling together and gaining consensus from the people in the various departments whose data, image files, sound, video or other types of content need to be assembled and categorized.

Not surprisingly, Verity specializes in software tools for mapping content into categories--what they call thematic mapping--and providing prebuilt taxonomies, usually build on an industry standard, such as the MeSH from Medline for medical information, or Factiva's company and industry taxonomy. Today's headline story about Verity's work on homeland security was referenced in the seminar, in large part because Verity had built a custom taxonomy in conjunction with the Department of Homeland Security for the project.

A couple of key take-aways from the seminar, which apply to both individual and institutional Web site publishers, as well as Web search engine companies:

--To achieve a quality result in creating and using a taxonomy to tag content, a combination of machine tools (i.e., software) and human review is necessary. One speaker put it this way: "people for quality; tools for quantity."

--A major benefit of having an underlying taxonomy is the ability to create browse categories that enhance text search. Note to search interface designers: user tests show that the best approach includes a text search window as a first step, followed by related browse categories that can be used to refine a search. Alternatively, navigation paths could be devised based on the browse categories that apply to the initial text search.

When the subject is taxonomies, it is easy to get bogged down in terminology: what is a taxonomy vs. a thesaurus vs. an ontology. The Verity speaker did a superb job of differentiating between the terms. I won't try to replicate her definitions. Instead, I'll close with this thought: As institutions increasingly apply system-wide taxonomies to their internal and externally-shared content collections, we will be well along the path toward the Semantic Web, where it will be far easier to determine context of individual sites on the Web and to identify communities of related content. Especially if some standards are adopted.