Re: need to determine what RDF is

From: "patrick hayes" <phayes@ai.uwf.edu>

> >Seth:
> >
> >I still fail to see why it is important for us to classify entailments.
It's
> >just going to over complicate stuff needlessly.  If my agent knows the
rules
> >for rdfs:subClass, than it can arrive at legitimate entailments,
> >
> >Pat:
> >
> >There is no universal overriding notion of 'legitimate' entailment,  is
why.
> >I agree it complicates things, but I see no way around the  fact that
life
> >is complicated.
> >
> >Seth:
> >
> >Life is complicated, agreed.  Consider the graph:
> >
> ><ex:Jane> <rdf:type> <ex:Woman>.
> ><ex:Woman> <rdfs:subClassOf> <ex:Human>.
> ><rdfs:subClassOf> <rdfs:comment>  "Indicates membership of a class".
> ><rdfs:subClassOf> <ex:rule>  "(=> (subClass ?SUBCLASS ?CLASS) (forall
> >(?INST)   (=> (rdf:type ?INST ?SUBCLASS) (rdf:type ?INST ?CLASS))))".
> >
> >Given that graph, do you agree that some agent could calucate that
> >"<ex:Jane> <rdf:type> <ex:Human>."   ??
>
> An RDFS-aware agent could, and that only requires the first two
> triples. The comments add nothing to the entailment.

Hmmm ... I doubt that any automated agent (whatever it advertises itself as)
will be able to calculate that without recourse to something like triple 4.

> >If so, then what kind of entailment is that, RDF, or RDFS, or is it EX,
or
> >is it RDF+EX+RDFS ?
>
> RDFS (which includes RDF).
>
> >  What are we to use that classification for?    In other
> >words what is the actual utility of that kind of thinking?
>
> The utility is that if I come across some published conclusions, I
> want to be able to find out what methods were used to arrive at those
> conclusions. If the thing that drew the conclusions advertises itself
> as an RDFS agent/reasoner, then I will be most upset if I find that
> it was using some inference method that is not RDFS valid. In fact, I
> might sue whoever wrote it for breach of contract and false
> advertising.

Hmmm ... i made the example too simple. Try this one:

[1] <ex:Jane> <rdf:type> <ex:Woman>.
[2] <ex:Woman> <rdfs:subClassOf> <ex:Human>.
[3] <rdfs:subClassOf> <rdfs:comment>  "Indicates membership of a class".
[4] <rdfs:subClassOf> <ex:rule>  "(=> (subClass ?SUBCLASS ?CLASS) (forall
(?INST)   (=> (rdf:type ?INST ?SUBCLASS) (rdf:type ?INST ?CLASS))))".
[5] <ex:Chiggy> <ex:rule> "(=> (rdf:type ?INST ex:Human) (rdf:type ?INST
ex:Chiggy))".
... and the entailment is ...
[6] <ex:Jane> <ex:type> <ex:Chiggy>.

Not the automated agent not only has to use rule [4] which it gets from the
legitimate RDFS MT, but it also has to use rule [5] from the namespace EX.
Now what kind of inference is it ?   Can you still sue?

Seth Russell

Received on Thursday, 30 May 2002 18:21:12 UTC