Re: deSkolemizing to get Horn-expressivity with RDF rules

From: Sandro Hawke (sandro@w3.org)
Date: 11/13/02

  • Next message: Mike Dean: "Joint Committee telecon tomorrow 19 November"
    > Well, ignoring the Skolem function issue for a moment, the 'existential-
    > conjunctive' power of RDF triples will also move RDF rules to a more
    > expressive position in the sublanguage hierarchy. But again, 'simulating'
    > Skolem functions by existentially quantified variables at the same time
    > puts us into a FOL with equality.
    
    How are you getting from RDF to disjunction and negation...?
    
    > > Puting it under
    > > datalog makes RDF rules fit nicely in your hierarchy, but it gives us
    > > a less powerful system than we (cwm/n3 users at least) want.
    > 
    > "RDF rules" is a much less well-defined term than "RDF triples", and in
    > RuleML we are happy to reclassify RDF rules into the most appropriate
    > position in the hierarchy once we understand them better.
    
    Good.  :-)
    
    > As I hinted at in a different thread, it should be helpful to exchange
    > our experience with rule engines such as cwm/n3 and Mandarax/RuleML or
    > j-DREW/RuleML.
    
    What would you suggest?   Our cwm/n3 work is all in a public cvs
    repository, but it hasn't been properly written up.  We're still
    trying to understand what aspects of it are novel and/or useful.   Do
    you have ideas for a process that might work here?
    
        -- sandro
    


    This archive was generated by hypermail 2.1.4 : 11/13/02 EST