Andy, Thx. Frank and I work together on this so all these notes had the same origin. To summarize:
1. it's reasonable to use rdfs:member as a predicate 2. the current ARQ behavior of eliminating rdfs:member triples is a bug 3. I can disable the behavior when using tdb.tdbquery either with an assembler or a Java wrapper - both require calling some Java code Regards, ___________________________________________________________________________ Arthur Ryman DE, PPM Chief Architect IBM Software, Rational Toronto Lab | +1-905-413-3077 Twitter | Facebook | YouTube From: Andy Seaborne <[email protected]> To: [email protected] Date: 06/15/2011 12:54 PM Subject: Re: ARQ Treatment of rdfs:member On 15/06/11 17:16, Arthur Ryman wrote: > I'd appreciate your views on an issue my dev org is running into. At > Rational we are planning to use rdfs:member as a generic membership > predicate, e.g. to indicate membership of resources in any set, not > necessarily an RDF Container. However, the default behavior of ARQ is to > compute rdfs:member using inference rules applied to the container > membership predicates rdf:_1, rdf:_2, etc. and it in fact ignores explicit > rdfs:member triples. > > RDF Schema inference rules should always add triples. However, the default > behavior of ARQ ignores explicit rdfs:member triples. This seems like a > bug. See reply to Frank. > I understand we can turn off this ARQ behavior in Java code. Is there a > way to do this using command line args to tdb.tdbquery? Not without getting it to call some java code, which is possible via assemblers initialization code. It's not exposes as a setting feature on the command line directly. One simple way for now though is to write a Java wrapper around tdb.tdbquery that calls ARQ.init(), makes the change, then calls tdb.tdbquery.main(...) > > What do you think is the intended use of rdfs:member, and how is it used > in practice? Thx. I've not seen it used explicitly but there is no reason why it can't be from SPARQL and ARQ point-of-view. Presumably, it just falls out in the rules anyway - it's just a triple is explicitly asserted and logically inferred. Andy > > Regards, > ___________________________________________________________________________ > > Arthur Ryman > > > DE, PPM Chief Architect > > IBM Software, Rational > > Toronto Lab | +1-905-413-3077 > Twitter | Facebook | YouTube > > > >
