Taxonomies are an integral element of many large law firms’ knowledge management programs. But building, maintaining, and using a taxonomy can be hard. 

In my recent post on Developments in Full Text Searching, I cited Sharon Flank, a computational linguist, of DataStrategy. Sharon has just released a white paper, Why Taxonomies are Doomed (PDF), that is worth reading. In it, she raises many of the issues that can make taxonomies troublesome.

I don’t necessarily agree with all her conclusions, especially with respect to the legal domain, but the issues she raises are certainly worth considering. My sense is that most large law firms do find taxonomies useful. The debate seems less one of “taxonomy or not” than one of “how deep should the taxonomy be.”

I have seen firms all over the spectrum on this latter point, some with very deep and complex taxonomies with hundreds of entries nested 5 or 6 levels deep, some with only a few tens of nodes in two levels. Partly for the reasons Sharon covers, I tend to lean toward keeping taxonomies simple. But I ultimately view this as an empirical question; the answer for any give firm will depend on its practices, its view of the world, how lawyers use PCs, the number of dedicated KM professionals, and a host of other factors.

One point that Sharon makes that seems especially apt for law firms is the importance of training users on how to search properly. A relatively small investment in the mechanics of searching can have a big pay-off. I recently had the chance to visit with a large firm where I was suprised to see that lawyers really understood how to use wild cards and Boolean techniques to find material of interest. One might think that is the norm, but in my overall experience, more lawyers need to know these techniques. If lawyers don’t know how to search, it’s not obvious to me that a sophisticated taxonomy will help them all that much.