Re: entailment review - part 1

On 6 Jan 2010, at 13:59, Birte Glimm wrote:

> 2010/1/5 Andy Seaborne <andy.seaborne@talis.com>:
> >
> > On 05/01/2010 7:43 PM, Birte Glimm wrote:
> > ...
> >>>
> >>> 4) "The term RDF-L denotes the set of all RDF Literals, RDF-B the set
> >>> of all blank nodes in RDF graphs"
> >>>
> >>> Hmmm, why "in RDF graphs" and in which graphs? BTW, this question also
> >>> applies to Query as well.
> >>
> >> That is more a query comment. I just repeat the query definitions as I
> >> also state above, to remind readers. If Andy and Steve want to change
> >> that, I happily use a changed definition.
> >
> > The reason is lost to me - RDF-B is only used to get to RDF-T anyway.
> >
> > (this used not to be true because at one time, you could have bnodes, as
> > non-distinguished variables, in the predicate position).
> 
> To me it seems to be unnecessary, but not really hurting on the other
> hand. Maybe it has historical reasons as Andy suggested.
> 
> > ...
> >
> >>> * Another remark which is not critical, but maybe we should re-discuss
> >>> it at some point ... BGP extension says that entailment regimes must
> >>> specify:
> >>>  - well-formed graphs
> >>>  - SG must be unitquely specified
> >>>  - entailment relation
> >>>  - finiteness condition for answers
> >>>  - handling of inconsistent graphs
> >>>
> >>> It doesn't *actually* say that it should define restrict "which qeries
> >>> are legal", does it? I anyway don't think that the definition of BGP
> >>> extension
> >>> does preclude such restrictions, but it isn't actually required by the
> >>> original definition.
> >>
> >> True. The closest to that is "An entailment regime specifies 1) ... 2)
> >> an entailment relation between subsets of well-formed graphs and
> >> well-formed graphs". and "2 -- For any basic graph pattern BGP and
> >> pattern instance mapping P, P(BGP) is well-formed for E". I am not
> >> sure whether I can interpret that as a possibility of defining what
> >> legal/supported queries are. I think I once discussed that with Andy
> >> and he suggested that all queries are legal, but some queries might
> >> have empty answers. In particular for OWL Direct Semantics, I would
> >> prefer to restrict not only the queried graphs but also the queries
> >> themselves. If a query BGP cannot be parsed into ontology structures
> >> then Direct Semantics entailment is just not defined. In that case I
> >> would prefer to raise an error instead of giving an empty answer.
> >> The other problem are update queries. Here we decided, I think, that
> >> we put a note somewhere that the entailment regimes document does not
> >> define the behaviour of systems for update queries. Once there is more
> >> implementation experience one can then specify what implemented
> >> systems do, which is most likely to use standard simple entailment for
> >> update queries. I can add a note in this direction.
> >>
> >>> 6) This remark might be overshooting (at leat for this WD), but:
> >>>
> >>> "The scoping graph, SG, corresponding to any consistent active graph
> >>> AG is uniquely specified and is E-equivalent to AG."
> >>> [...]
> >>> "All entailment regimes specified here use the same definition of a
> >>> scoping graph as given in SPARQL 1.0. Thus, the required equivalence
> >>> is immediate."
> >>>
> >>> I am a bit worried that *actually* the definition of the scoping graph
> >>> as given in SPARQL 1.0 is *NOT* uniquely specified, since it obviously
> >>> doesn't
> >>> uniquely determine the blank nodes. Not sure whether this is really an
> >>> issue, but it seems a bit awkward.
> >>>
> >>> Maybe the condition should be weakened to something like
> >>>
> >>> "The scoping graph, SG, corresponding to any consistent active graph
> >>> AG is uniquely (except blank node identifiers) specified and is
> >>> E-equivalent to AG."
> >
> > IIRC It's not supposed to uniquely specify it in the query spec but to give
> > the framework - the entailment regime should specify the scoping graph in a
> > compatible manner.
> 
> I agree that the Query spec says an extension to BGP matching has to
> uniquely specify the SG. That seems to me a bit over the top or maybe
> I am interpreting that too strictly? You have to pick one of a number
> of equivalent scoping graphs. The ones you can choose from are the
> ones that differ in bnode names and the only requirement is that
> bnodes are suitably named apart, i.e., the query and the SG should not
> share any bnodes. I would just leave it as that unless there are
> complaints. 

Well, this actually *is* sort of a complaint. I admittedly would feel safer, 
if the condition was changed to something like the suggested:

 "The scoping graph, SG, corresponding to any consistent active graph
  AG is uniquely (except blank node identifiers) specified and is
  E-equivalent to AG."

or

 "The scoping graph, SG, corresponding to any consistent active graph
  AG is uniquely (modulo equivalence) specified and is E-equivalent to AG."

Would that be acceptable?

> It would be hard to specify it so that it is more unique,
> not knowing the given inputs. The result would not differ either way.

yes, but leaving it as it stands now requires uniqueness, which is in 
my opinion too strict.

> 
> >>> Not really ideal either, but better than before?
> >>> If we agree on that change, we can include that with a remark to ask
> >>> for comments?
> >>
> >> That is again a comment for Query and I agree it is a valid comment.
> >> Several of the given conditions/definitions are not ideal IMO, that
> >> being one of them. I would also prefer to use a skolemized scoping
> >> graph directly, but that is also not possible, so I define this kind
> >> of work around to meet the Query conditions. We further violate
> >> already against the condition that the scoping graph must be
> >> consistent according to the conditions in the Query spec, which we
> >> cannot guarantee with the current RDFS entailment regime definition. I
> >> would prefer to be more consistent, i.e., either remove the
> >> consistency requirement everywhere or have it throughout.
> >
> > I think we have more latitude with the defns here because they exist for the
> > purposes of extension and so are less tested by the spec as published as the
> > REC.  I don't have the bandwidth to work on it before the upcoming
> > publications.  We need wider review than just the WG can currently manage -
> > maybe we need to seek out explicit reviews.
> 
> What I could do is to put in an editorial note into the entailment
> regimes doc that mentions that the SG is only unique up to bnode names
> and that the SG is not necessarily consistent (well that's already in
> an ed note) explicitly asking for feedback on whether that is seen as
> problematic given the requirements from the Query spec. Is that a good
> idea?

Sure, an ed note would make me happy for now, if we can't reach agreement 
on that suggested change.

best,
Axel

> Birte
> 
> >        Andy
> >
> 
> 
> 
> --
> Dr. Birte Glimm, Room 306
> Computing Laboratory
> Parks Road
> Oxford
> OX1 3QD
> United Kingdom
> +44 (0)1865 283529
> 

Received on Wednesday, 6 January 2010 14:19:19 UTC