Re: time ontology should allow "conference1 after conference2"

On Tue, 2006-02-21 at 19:24 -0500, Dan Brickley wrote:
> * Dan Connolly <connolly@w3.org> [2006-02-21 18:05-0600]
> > 
> > I just found
> >  http://www.isi.edu/~pan/SWBP/time-ontology-note/time-ontology-note.html
> > via WG minutes and such.
> > 
> > :TemporalEntity
> >       a       owl:Class ;
> >       rdfs:subClassOf :TemporalThing ;
> >       owl:equivalentClass
> >               [ a       owl:Class ;
> >                 owl:unionOf (:Instant :Interval)
> >               ] .
> > 
> > and the domain/range of intBefore is constrained to TemporalEntity.
> > So I can't just say
> >   :conference1 swbp-time:intBefore :conference2.
> > without implying that :conference1 is an :Instant or :Interval.
> > I have to have some property that relates a conference to
> > a time interval, and then use :intBefore on that.
> > 
> > I can't see any reason for the indirection.
> > 
> > The cyc ontology seems to express all the relevant stuff
> > without this constraint. The cyc after relationship
> > applies to not only instants and intervals, but also
> > conferences, meetings, people, etc.
> > 
> > http://www.cyc.com/cycdoc/vocab/time-vocab.html#after
> 
> I can imagine that 'afterness' might mean slightly different things
> when relating different types of entity. So doing away with the
> indirection is tempting. But sometimes a thing is a member of multiple
> independent classes; eg. be both an xyz:Person, an abc:Employee and 
> cde:RFIDTaggedEntity. Since we can't rely on typing information always
> being present, this makes me wary of overloading 'after', since 
> multiple class membership could bring different interpretations of 
> afterness to bear.

I don't see how multiple types or lack of information is relevant.
It's constraints that lead to inconsistencies that I want to avoid.

> 
> That said, from the definition at 
> http://www.cyc.com/cycdoc/vocab/time-vocab.html#after I don't see 
> how it applies to people, meetings and random other types, 
> "A #$PrimitiveTemporalPredicate  that relates two points in time.
> (#$after LATER EARLIER) means #$TimePoint  LATER is after (occurs later
> in time than) #$TimePoint EARLIER."
> Does some other bit of Cyc model people, conferences etc as 'time
> points'? Or am I missing something obvious?

oops; wrong one.

The ones I tend to use are...

http://www.cyc.com/cycdoc/vocab/time-vocab.html#$startsDuring
http://www.cyc.com/cycdoc/vocab/time-vocab.html#temporallySubsumes
http://www.cyc.com/cycdoc/vocab/time-vocab.html#temporalBoundsContain

hmm... those seem to be defined in terms of after and
http://www.cyc.com/cycdoc/vocab/time-vocab.html#startingPoint

e.g.
http://www.cyc.com/cycdoc/vocab/time-vocab.html#$startsAfterEndingOf
"Thus it is equivalent to the form (#$after (#$StartFn AFTER) (#$EndFn
BEFORE))."

so I suppose the cyc design uses an indirection too; but the
TemporalThing class is _not_ constrained to be an instant
or interval.

http://www.cyc.com/cycdoc/vocab/time-vocab.html#TemporalThing


> 
> Dan
-- 
Dan Connolly, W3C http://www.w3.org/People/Connolly/
D3C2 887B 0F92 6005 C541  0875 0F91 96DE 6E52 C29E

Received on Wednesday, 22 February 2006 13:31:14 UTC