RE: data smushing

Jos,

Yes, I think this is a central technique.

I think what you describe is the broader aspect of inference, rather than 
unification which is one specific technique of some inference 
engines.  (But I am not an expert here!)

#g
--

At 11:45 AM 1/4/01 +0100, you wrote:
> > > > If
> > > >    [nodeX, propertyY, whatever] and;
> > > >    [nodeY, propertyY, whatever] and;
> > > >    [nodeX, rdf:type, nodeZ] and;
> > > >    [nodeY, rdf:type, nodeZ] and;
> > > >    [propertyY, atMostOneEntityValue, "yes"];
> > > > then
> > > >    smush (nodeX, nodeY).
> > >
> > >
> > >Isn't 'smushing' just unification hacking; am I missing something?
> >
> > That's an interesting thought.
> >
> > I think, however, that there's more to smushing (which I understand to mean
> > detection of equivalent resources from their description and/or usage).
> >
> > Unification uses a sequence of variable->subexpression substitutions to
> > make two expressions the same, and is based entirely on the form of the
> > expressions concerned.
>
>I think we just look to the same thing from different perspectives.
>Suppose we have the facts
>{Maaike|ed:human|}   i.e. Maaike is a human
>{Maaike|ed:female|}  i.e. Maaike is a female
>{Goedele|ed:woman|}  i.e. Goedele is a woman
>
>then, because "human and female defines woman", we would like to detect
>{Maaike|ed:woman|}
>
>One can of course use different mechanisms to find that "smushed data".
>One possible way is using an explicit description of what we accept
>{{?X|ed:human|}|e:impliedBy|
>   {?X|ed:woman|}}
>
>{{?X|ed:female|}|e:impliedBy|
>   {?X|ed:woman|}}
>
>{{?X|ed:woman|}|e:impliedBy|
>   {?X|ed:human|}
>   {?X|ed:female|}}
>
>and so on ...
>to find a proof
>{{Maaike|ed:woman|}|e:impliedBy|
>   {Maaike|ed:human|}
>   {Maaike|ed:female|}}
>
>which can be passed as evidence.
>
>--
>Jos De Roo, AGFA http://www.agfa.com/w3c/jdroo/

------------
Graham Klyne
(GK@ACM.ORG)

Received on Thursday, 4 January 2001 07:58:20 UTC