Re: New DQL Specification

From: Peter F. Patel-Schneider (pfps@research.bell-labs.com)
Date: 05/29/02

  • Next message: Mike Dean: "Joint Committee telecon tomorrow 4 June" 
    From: patrick hayes <phayes@ai.uwf.edu>
Subject: Re: New DQL Specification
Date: Tue, 28 May 2002 16:40:12 -0500

> >More problems with MIDs.
> >
> >What happens in the following cases (some written in quasi English, because
> >the DAML+OIL would be too painful):
> >
> >
> >KB	John rdf:type _:r .
> >	_:r daml:onProperty friend .
> >	_:r daml:minCardinality 3 .
> >
> >Query	John friend ?l .
> >	?l distinguished
> 
> no answer binding.
> 
> >KB	John rdf:type _:r .
> >	_:r daml:onProperty friend .
> >	_:r daml:minCardinality 3 .
> >	John friend _:f .
> >
> >Query	John friend ?l .
> >	?l distinguished
> 
> ?l/_:f ; John friend _:f .
> 
> Since the MID in this case is exactly the query, it might be worth 
> not delivering it as part of the answer. Similarly for later examples.

In the above two cases the KBs are equivalent.  It seems exceedingly
strange that two equivalent KBs can give rise to different answers to a
query.  I view this a fatal problem that must be solved.

> 
> >KB	John rdf:type _:r .
> >	_:r daml:onProperty friend .
> >	_:r daml:minCardinality 3 .
> >	John friend _:f .
> >	John friend _:g .
> >
> >Query	John friend ?l .
> >	?l distinguished
> 
> ?l/_:f ; John friend _:f .
> ?l/_:g; John friend _:g .

How can you force two answers, as either answer ``covers'' both _:f and
_:g?

> >KB	John rdf:type _:r .
> >	_:r daml:onProperty friend .
> >	_:r daml:toClass Person .
> >	John rdf:type _:s .
> >	_:s daml:onProperty friend .
> >	_:s daml:minCardinality 1 .
> >
> >Query	John friend ?l .
> >	?l distinguished
> 
> No answer binding.

Again, 

       John fried _:f .

is a consequence of this KB, but you would have a different answer if this
triple is added.

> >KB	John rdf:type [all friend Student OR all friend Employee ]
> >	John friend _:f .
> >
> >Query	John friend ?l .
> >	?l distinguished
> 
> ?l/_:f; Im not sure what the MID would be here; it depends to some 
> extent on how that OR would be encoded in the RDF graph. Thinking 
> about it, I think that what you would get would be a graph which when 
> back-translated into this notation would look like this:
> 
> ?l/_:f; _:f rdf:type [Student OR Employee]

Well, your definition of the MID would not get there, as there is no
connection from _:f to [Student OR Employee] in the RDF graph of the KB.
In fact [Student OR Employee] is not even in the RDF graph at all.

> >KB	John rdf:type [all friend Student OR all friend Employee ]
> >	John friend _:f .
> >	John friend _:g .
> >
> >Query	John friend ?l .
> >	?l distinguished
> 
> ?l/_:f;  _:f rdf:type [Student OR Employee]
> ?l/_:g;  _:g rdf:type [Student OR Employee]
> 
> is about the best one could expect.

See above.

> >KB	John rdf:type [all friend Student OR all friend Employee ]
> >	_:r daml:onProperty friend .
> >	_:r daml:minCardinality 2 .
> >
> >Query	John friend ?l .
> >	?l distinguished
> 
> No answer binding.

This KB is equivalent to the previous one.

> >KB	John rdf:type [friend _:f OR friend _:g ]
> >
> >Query	John friend ?l .
> >	?l distinguished
> 
> No answer binding.

Again, there is an equivalent KB for which you would return an answer binding.

> Pat


All this brings up the basic point:

    What is the definition of an answer?

I don't think that this question has been satisfactorily answered.

peter

    This archive was generated by hypermail 2.1.4 : 05/29/02 EDT