11 October 2004

More on the Semantic Web

WebReference.com has a new article on the semantic Web.

The most compelling writing on the semantic Web includes hypothetical applications. It-feels-right when I hear these fanciful scenarios of emergent, intelligent application. Such benefits mimic those of well-designed software: through rigorous adherence to programming principles within the architecture, features can emerge with less effort. Scenarios one and two contain some stiff but colorful narratives. The key concept in those narratives is intelligent agents--a semantically described World Wide Web would make agents easier to program. They would emerge more easily from the manner in which content is described.

Section 1.4 shows Tim Berners-Lee's semantic Web layer-cake, stolen here for your benefit:

semantic Web layer-cake

The most interesting layers, linguistically, are the "Logic and Proof" and "Ontology" layers, containing respectively the predicate calculus [Wikipedia] used to derive truths and the semantic definitions used within the logical equations. Equations and variables.

It may seem a little pie-in-the-sky, but it's actually more reasonable than most natural language problems. Chatterbots [Wikipedia], when given a wide domain, are still very feeble. The structure of the semantic Web would limit that domain and provide simple rules for extending the knowledge.

[ posted by sstrader on 11 October 2004 at 9:04:24 PM in Programming ]