The Taxonomy Torpedo

March 9, 2010

Quite an interesting phone call today (March 8, 2010). Apparently the article “A Guide to Developing Taxonomies for Effective Data Management” caught this person’s attention. The write up boils down the taxonomy job to a couple of pages of tips and observations. Baloney in my opinion.

The caller wanted to have me and a gosling provide the green light to a taxonomy project. The method was to use a couple of subject matter experts from marketing and an information technology intern. The idea was to take a word list and use it to index content with the organization’s enterprise search system.

The called told me, “We let the staff add their own key words. There has been a lot of inconsistency. We will develop our controlled term list and that way we have date, time, and creator; the terms the users assign; and the words in our taxonomy. What do you think?”

What I think is that no one will be able to find some of the relevant data. I am surprised that so many vendors point out that their systems “discover” metadata and provide users with suggestions, lists of related content, and the ability to search by entities.

Doesn’t work.

Here’s why:

  1. Fancy interfaces (user experience in today’s lingo) requires consistent, appropriate, and known tags. Most organizations, fresh from doing taxonomy push ups for a day, have wildly inconsistent term lists. A user may know how to locate a document in an idiosyncratic way. If that method involves a controlled word, the user may not get the results she was expecting.
  2. Automatic processes work well when the information objects have enough substantive content to make key word indexing work. I have examined a number of organizations’ content and found inconsistencies in the way in which the organization referred to itself. The controlled terms were rarely used. When a query included a controlled term, the user was puzzled why the result set was not complete.
  3. Most organizations lack the expertise and resources to create a well-formed controlled term list. Ad hoc lists are useful sometimes to those who cooked them up. A comprehensive controlled term list is a great deal of work.

What’s this mean? The stampede to taxonomies will yield the same dissatisfaction that other, partially implemented search features. Talk is easy. Taxonomies and controlled term lists are tough to develop and even harder to keep current.

Stephen E Arnold, March 9, 2010

No one paid me to write this. I mention indexing, so I will report non payment to the Librarian of Congress, or maybe the librarian for the House library, or maybe the librarian for the Senate library. I wonder why there are three libraries for Congress.


One Response to “The Taxonomy Torpedo”

  1. The Accidental Taxonomist | Digital Asset Management on April 21st, 2010 1:56 am

    […] The Taxonomy Torpedo ( […]

  • Archives

  • Recent Posts

  • Meta