From: pat hayes (phayes@ai.uwf.edu)
Date: 06/11/01
>On June 5, Peter F. Patel-Schneider writes: > > From: Frank van Harmelen <Frank.van.Harmelen@cs.vu.nl> > > Subject: Re: rule proto-proposal > > Date: Tue, 05 Jun 2001 19:38:34 +0200 > > > > > I wrote: > > > > > > > >Instead, we are designing a language > > > > >1. that draws conclusions (and not one that makes transformations), > > > > >2. for DAML+OIL (and not for arbitrary RDF). > > > > >that is: rules which logical conclusions that are sanctioned by the > > > > >semantics of DAML+OIL. > > > > why designing a language and cripple the scope when the >broader language > > > > can do everything? > > > > > > Stefan replied (in an off-line email): > > > > > > > Transforming RDF provides the means for manipulating languages. > > > > Of course the rules that manipulate have to respect the >semantics of the > > > > languages, but thats the responsibility of the rules author. > > > > > > BIG DISAGREEMENT! I would expect DAML+OIL rules always to >respect DAML+OIL semantics, ie: if S is some statement which I hold >true, and R is a rule which I hold true, than the result of applying >R to S is a conclusion which I must unavoidably regard as true. > > > > > > I wouldn't want DAML+OIL rules to be so flexible that authors >could write rules for which the above didn't hold. If DAML+OIL rules >would be "just" RDF transformation rules, authors could write such >nonsensical (non-semantics respecting) rules. > > > > > > Stefan added: > > > > > > > Not doing what I propose means the DAML+OIL rule language would be > > > > superfluous very shortly. > > > > > > I think an RDF transformation rule and a DAML+OIL inference >rule are two very different animals. We should be clear on which one >we are breeding. > > > > > > Frank. > > > ---- > > > > I agree with Frank on this issue. > >I would like to add my agreement to the record. > >Ian This is a bit like trying to make peace in the middle east, but let me suggest that it might help if we simply changed the terminology, since "rules" means different things to different people. Lets call what Stefan is proposing not a 'rule' language but a transformation language (TL) and lets call a language which extends the expressive power of DAML while preserving the model theory an extended inference language (EIL). Peter and Frank and Ian are most interested in an EIL. Stefan thinks the world urgently needs a TL , and I think he is right; and he also thinks that a useful EIL might be implemented as a restricted form of a logic-based TL, killing two birds with one stone, as it were; which I am less confident about but would be delighted to see done if it could be done. But in any case, in the short-to-medium term, it seems more useful to keep these two goals - the EIL and the TL - as separate goals, and to develop them independently, rather than arguing about which is best. They both will have their uses, I am sure. My original strawman, and Peter's, were both intended to be EILs. They differ dramatically, so one of us must be wrong. Any technical views on their relative merits? Pat Hayes --------------------------------------------------------------------- IHMC (850)434 8903 home 40 South Alcaniz St. (850)202 4416 office Pensacola, FL 32501 (850)202 4440 fax phayes@ai.uwf.edu http://www.coginst.uwf.edu/~phayes
This archive was generated by hypermail 2.1.4 : 04/02/02 EST