[Developers] Reasoner?

[Peter Burns]

Is the plan to support transitivity in a generalized way?  Could someone
assert that ancestry is transitive in such a way that a query asking "is
Aaron Burr an ancestor of Jody Foster" would cause metaweb to walk up
'parent' links and give an answer?


On 10/2/07, Robert Cook <robert at metaweb.com> wrote:
>
> Just so you know, it's one of our goals to support transitive closure in
> particular on geographic containment and do it in a performant way.  That
> is, your French University example below should work with little penalty in
> execution time.  I don't currently have an ETA, but it's something we're
> serious about.
>
> R
>
> ----- Original Message -----
> From: "Kavitha Srinivas" <ksrinivs at gmail.com>
> To: "Shawn Simister" <narphorium at gmail.com>
> Cc: developers at freebase.com
> Sent: Tuesday, October 2, 2007 7:21:02 PM (GMT-0800) America/Los_Angeles
> Subject: Re: [Developers] Reasoner?
>
> Shawn has a good point.  If you want to describe role hierarchies, or
> transitivity (e.g., locatedIn is transitive), then you can use OWL
> and use OWL reasoners to do this.
> Kavitha
>
> On Oct 2, 2007, at 6:08 PM, Shawn Simister wrote:
>
> > Once the Freebase data is available as RDF it will be possible to
> > reason
> > about it using any of the existing RDF reasoners out there. I'm
> > working
> > on making this happen but I only have a little bit of free time
> > right now.
> >
> > Shawn
> >
> > Christoph Pingel wrote:
> >> Hello,
> >>
> >> I'm wondering if there is any artificial reasoner involved in
> >> freebase? My impression from looking into MQL is that queries only
> >> result in the explicit structures, not implicit ones.
> >>
> >> For example, can I ask for universities in France and get *all*
> >> French universities even if "contained by Lyon" or "contained by
> >> Toulouse" are the only location values that are there?
> >>
> >> Another case would be the research for people with a certain
> >> expertise that can be expressed in rules like
> >>      "person knows about topic if
> >>              person wrote book about topic
> >>              or person wrote article about topic
> >>              or person organized workshop about topic
> >>              or person curated exposition about topic
> >>      ".
> >>
> >> My second question is somewhat related. Perhaps it would make sense
> >> to have specialisation hierarchies (or inheritance chains) for
> >> relationships as well, not just for types.
> >>
> >> Looking at movies, for example, all the actors, the director etc. are
> >> all involved with a certain movie. From time to time, there are cases
> >> when it would be convenient to just ask for »people involved«...
> >> Hm, I'm not sure. Perhaps what I'm asking for is to have inheritance
> >> between compound value types...
> >>
> >> best regards,
> >> Christoph
> >> _______________________________________________
> >> Developers mailing list
> >> Developers at freebase.com
> >> http://lists.freebase.com/mailman/listinfo/developers
> >>
> >>
> >
> > _______________________________________________
> > Developers mailing list
> > Developers at freebase.com
> > http://lists.freebase.com/mailman/listinfo/developers
>
> _______________________________________________
> Developers mailing list
> Developers at freebase.com
> http://lists.freebase.com/mailman/listinfo/developers
>
> _______________________________________________
> Developers mailing list
> Developers at freebase.com
> http://lists.freebase.com/mailman/listinfo/developers
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: http://lists.freebase.com/pipermail/developers/attachments/20071002/e2d57e3c/attachment.htm