The Knowledge Management Connection   

Powerful full-text search (FTS) is vital to any system for retrieving content. But full-text search by itself has serious flaws:

• You simply may not know the name of what you are looking for.
• Misspellings are poisonous to full-text queries.
• You may not use the same vocabulary as the author.
• An excessive number of hits in retrieval (inadequate precision) substantially increases the effort of searching.
• The elements in a text-only database are disassociated. You may find one highly relevant unit, but how do you find closely related information?

In spite of the remarkable improvements in full-text search in the last 20 years, these shortcomings of full-text retrieval engines persist, and they can make the difference between finding critical information quickly and not finding it at all.

Those are the technical difficulties of full-text search. But consider also the things that FTS simply was not designed to do:

• Identify what’s important.
• Identify relationships among concepts.
• Identify differences between the outside world’s knowledge and the special knowledge of the organization.
• Capture emergent differences — new concepts and changes in existing concepts.
• Identify people and associate them with topics.
• Kill outdated and erroneous information.
• Differentiate organizational terminology from the outside world's terminology.
• Identify what’s irrelevant.
• Establish ownership, authority, and approval.

The alternative is to classify the information and to associate those terms with both documents and subject-matter experts. Because even if you convert a major collection of “documents” into text form, only a systematic classification of knowledge resources will dramatically increase your chances of successful retrieval.

Imagine how difficult it would be to find information on a specific topic in an online text version of all the content of all publications in the Boston Public Library … without a card catalog, electronic or otherwise.