Friday, February 15, 2008

The Semantic Web - Not Quite Machines Only

The semantic web was designed to allow machines to function on the web independent of human beings. Semantic web ontologies provide definitions for terms that machines can understand, but if a machine needs a definition for something that it has encountered how does it find a document with this term?

UMBC's Swoogle is a semantic web search engine designed to retrieve semantic web documents for queried terms. Yet even if a robot could go to Swoogle and retrieve a set of semantic web documents how does it know which one has the best definition for the term it needs?

I've been working on a way to solve this problem using standard AI state space search algorithms to find the best definition for a term based upon terms it occurs with. Since you need definitions for all terms you should try to find ontologies which define as many of your terms as possible that way you can cover the terms with the fewest semantic web documents.

No comments: