Magic May Not Come From Pre-Made Taxonomies

June 17, 2015

There are hundreds of companies that advertise they can increase information access, retrieval and accuracy for enterprise search by selling prefabricated taxonomies.  These taxonomies are industry specific and are generated by using an all-or-nothing approach, rather than individualizing them for each enterprise search client.  It turns out that the prefabricated taxonomies are not guaranteed to help enterprise search; in fact, they might be a waste of money.  The APQC Blog posted “Make Enterprise Search Magical Without Money” that uses an infographic to explain how organizations can improve their enterprise search without spending a cent.

APQC found that “best-practice organizations don’t have significantly better search technology.  Instead, they meet employees’ search needs with superior processes and approaches the content management.”

How can it be done?

The three steps are quite simple:

  1. Build taxonomies that reflect how people actually think and work-this can be done with focus groups and periodically reviewing taxonomies and metadata. This contributes to better and more effective content management.
  2. Use scope, metadata, and manual curation to ensure search returns the most relevant results-constantly the taxonomies for ways to improve and how users are actually users search.
  3. Clear out outdated, irrelevant, and duplicate content that’s cluttering up your search results-keep taxonomies updated so they continue to deliver accurate results.

These are really simple editing steps, but the main problem organizations might have is actually implementing the steps.  Will they assign the taxonomy creation task to the IT department or information professionals?  Who will be responsible for setting up focus groups and monitoring usage?  Yes, it is easy to do, but it takes a lot of time.

Whitney Grace, June 17, 2015

Sponsored by ArnoldIT.com, publisher of the CyberOSINT monograph

 

Comments

Comments are closed.

  • Archives

  • Recent Posts

  • Meta