Article:		The Semantic Web is Closer Than You Think
Subject:		what's the point?
Date:		2003-08-21 10:22:21
From:		Bill Seitz
When I hear about machines reasoning about domains, my AI-skeptic agent cries for its momma. Is there a document out there sketching a useful/attractive scenario that could be achieved via a reasonable process of decentralized adoption over the next 3 years?

Previous Message

Next Message

• what's the point?
  2003-08-22 06:59:37 Sebastian Schaffert [Reply]
  
  The point is, that OWL builds on Description Logics which is a restriction of first order logic in which everything is decidable (typically in PSPACE). In this way, many of the problems of classic AI can be avoided, at the cost of a possibly reduced expressivity.

  
  Also, classic AI often tried to find a solution for everything, e.g. a "world ontology", which is an almost impossible task. The approach currently taken, however, is to use much smaller ontologies that are domain specific (e.g. specific to libraries or specific to geography).

  
  However, developing even such small ontologies will still be a major effort, and I think that not many people are capable to design proper ontologies even for comparably small fields.

  
  Anyway I think this is a step forward (even if it is small) and I think it will get some acceptance. OWL is part of the "RDF done right" work, now that many AI people (e.g. Pat Hayes) are working on the Semantic Web.

  
  Sebastian
  
  

  • what's the point?
    2003-08-26 20:29:26 Adrian Edwards [Reply]
    
    ... and then the thory is that all of these "small ontologies" will be able to interoperate on shared concepts, leveraging the "web effect" to grow rapidly into a large ontology approaching the kind of breadth needed for really useful reasoning.

    
    Note: The basis for interoperability of OWL ontologies is exactly the same as the basis for linking all the HTML pages on the web: the URI.