Semantic Networks

James Carlyle writes “I’ve recently come across three semantic networks which might be of interest to XML developers. A semantic network is a series of links between senses of words, with the links being hypernyms, synonyms etc. 1. Signiform’s ThoughtTreasure, which contains 25,000 concepts, 55,000 English and French words and phrases and 50,000 assertions. 2. Princeton University’s WordNet is far more ambitious and older, and contains 173,941 senses. An XML interface to WordNet has been put together using RDF by Dan Brickley, of the W3C, and is available at xmlns.com 3. Cyc is the most famous (or infamous) ontology project, still going after 15 years. Very comprehensive upper-level but very confusing even to so-called “ontologists”. Dan Brickley’s WordNet work is of interest, since he proposes that URIs of specific senses of words could be used to disambiguate meaning, so we could be sure that the <port> element in your document had the same meaning as mine. In addition, if your document used the element <port> and my processing expected the element <harbour>, WordNet could allow the processor to make a lateral interpretation.”

It’s work noting that Cyc is being trialled at HotBot. Search for something like Java and HotBot will suggest different senses of the word Java: Java (island), Java (computer language), coffee (drink). These link straight back into the search engine as better refined search. So clicking on Java (computer language) reformulates the search as Java “computer language” in HotBot. It helps inexperienced search engine users build better queries. Now which enterprising developer is going to create a meta search engine that uses WordNet to create more specific searches for a whole range of search engines?

Permalink: http://blog.iandavis.com/2000/02/semantic-networks/

Other posts tagged as semantic-web

Earlier Posts