_Latent Sematic indexing (LSI) Content, anyone heard about this?

NLP in this context is Natural Language Processing

 

Well, these days anyone can attach the "semantics" token to label a product or service and next thing one hears is a claim for LSI, semantics, or NLP.

Regarding Hakia, I would like to see better results coming from this "semantic" tool that deploys results from ASK --as they state. Results from basic tests are not encouraging.

Most search engines cannot discriminate between queries with same tokens, but different meanings (polysem semantics).

For example, a test for college junior and junior college in the defaul mode (AND in most engines) in Google, ASK, and Hakia are not encouraging.

Results in EXACT in Google are a bit better, though, but not in Hakia.

When reformulated as questions using:

Who is the best college junior?
Who is the best junior college?

(or reformulated using Who, Where, Which, What, etc)

in Hakia shows this engine still do not understand the semantic information need asked.

The case of polysemy with single terms is not encouraging, either.

This is not unique to Hakia as most current engines are a long shot at understanding meaning upon an information need sought.

 

Listen to orion!