Tom Reamy is, to many of us, one of the authorities on taxonomies. He’s talking right now about the varieties of taxonomy/text analystics software available now, and how to choose which is best for your application & organization. tom-reamy.jpg  All the vendors are adding more text analytics. So…step number one is: how are you going to use the technology?  Text mining? business inteligence? cusomter intelligence? tfor facted navigation? keyword indexing? to browse the front end of the portal?

Evaluating Taxonomy Software:

  • new, copy, rename, delee, merge
  • scop notes
  • spell check
  • search
  • names & identifiers
  • versioning
  • ease of use
  • user documentation
  • visualiztion — how does it show things?
  • automatic taxonomy/node generation — Tom says it’s nonsense but can be used at different stages, especially to get suggestions within a node or for entity extraction
  • entity extraction
  • auto-catgorization (training sets, terms, rules, advanced – saved search queries) “near” sentence” “paragraph”
  • boolean search that allows you to search for x near y, and “not”
  • advanced features — sentiment analysis (for customer service to see what people are saying about them); facts, ontologies, semantic web, etc…..

Phew!  If you want taxonomy management only, you are probably in a small company with a specialized taxonomy – and the good news is that this type of software is quite affordable. But do check the upgrade path for this type of tool, just in case you need to grow the application.Advanced application platforms, sich as Attensity’s or Inxight’s, are for those appls that need to integrate search and content management or to integrate policy, procedures and distributed contributions.

Tom’s advice: forget score cards to evaluate & do a pilot project instead  — use yoru content, in your application to see HOW the tool really works— “think big, start small & win”

The “C” level problem — is that someone at the CEO, CFO, COO level has to approve this, & they won’t know a thing about it; all you can do is demonstrate the complexity in their language — tell them stories of the pain they recognize in the organization.

Key ingredient to evaluating:

  • start with your own self knowledge – understand your content as much as you can, the technology already in place, and the business & information behaviours of the people that will be using the applications
  • eliminate the unfit according to the list above — ask experts, look at reputation of vendors, make sure the tool matches your scope & your environment — narrow it to 3-4 vendors, bring them in, have them do demos with your content, have them all in on one day
  • deep pilot — how well does it work with the semantics? this gives you a much better handle on making the decision between your top 2 vendors