Re: Summary of recent issues on current query spec (to be discussed in detail after current WD round)

> 1) consistency requirement for entailment regimes
>
>  Issue text: For efficient implementations, it might be undesirable to enforce
>  consistency checking, e.g. for RDFS. So, in order to follow:
>  "The effect of a query on an inconsistent graph is not covered by this specification,
>  but must be specified by the particular SPARQL extension."
>  My (Axel- chairthatoff) interpretation is that I don't see that this implies that an
>  extension has to uniquely define the behavior on
>  inconsistent graphs, actually it could leave several options open (e.g. for implementations that
>  do or don't perform consistency checking.)

The problem with the current definition is that it does not guarantee
finite answers in the case of inconsistencies. This is because SG is
defined in any case as equivalent to AG, which results in SG being
inconsistent. Now answers are those that are entailed by SG and
satisfy the extra conditions, which limit infinite answers from blank
nodes and axiomatic triples. An inconsistent graph, however, entails
everything and things such as:
ex:a ex:b "<a1>"^^rdf:XMLLiteral .
ex:a ex:b "<a2>"^^rdf:XMLLiteral .
...
ex:a ex:b "<an>"^^rdf:XMLLiteral
are all valid answers not filtered out by the current conditions. The
conditions were defined with the handling of inconsistent graphs by
current systems in mind. Current system behaviour can be captured by
saying that in case of an inconsistent AG, SG is defined as a
"repaired" AG where malformed XML literals are replaced by arbitrary
wellformed ones. With that definition of SG finite answers are
guaranteed and systems can keep doing what they do. Another way, which
I am not sure whether that satisfies the conditions on entailment
regimes is to say that in case AG is inconsistent the system behavior
is not defined precisely by the spec and systems that don't check
consistency are required to make sure that they always return finite
answers in a system specific way. This might be the better option
since repairing RDF graphs will no longer work for OWL Full since
there are many reasons for inconsistencies and many ways of repairing
(not just replacing malformed literals).
A third way is to tighten conditions on answers in general and require
that not only subject bindings occur in the input vocabulary, but also
the bindings for predicates and objects.


> 2) uniqueness of scoping graph
>
>  The current spec of extending BGP matching requires uniqueness of the scoping graph, whereas actually the definition
>  of a scoping graph for simple entailment is only unique up to homomorphic (i.e. simple) equivalence:
>
>  "1 -- The scoping graph, SG, corresponding to any consistent active graph AG is uniquely specified and is E-equivalent to AG."
>
>  It is there fore discussed to clarify this condition, e.g.:
>
>  "The scoping graph, SG, corresponding to any consistent active graph
>  AG is uniquely (modulo simple equivalence) specified and is E-equivalent to AG."

The query spec claims that simple entailment satisfies these
conditions, which is also only true up to bnode renaming:
"It is straightforward to show that SPARQL satisfies these conditions
for the case where E is simple entailment, given that the SPARQL
condition on SG is that it is graph-equivalent to AG but shares no
blank nodes with AG or BGP (which satisfies the first condition). "
and
"... pattern solutions are therefore understood to be not from the
active graph of DS itself, but from an RDF graph, called the scoping
graph, which is graph-equivalent to the active graph of DS but shares
no blank nodes with DS or with BGP"
Thus also simple entailment just takes any graph-equivalent scoping
graph, which does not give a *unique* definition for the SG. Changing
the spec in this regard seems also necessary for simple entailment.


> 3) Definition of RDF-B
>
>  ""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? It might be clearer/easier to simply drop "in RDF graphs",
> or to specify which graphs are talked about here.

Agreed.

> 4) Definition of Pattern Instance Mapping
>
>  Birte's suggested clarafication, cf. http://lists.w3.org/Archives/Public/public-rdf-dawg/2010JanMar/0029.html

Andy, what is a mapping pattern?

Birte

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

Received on Monday, 8 February 2010 13:50:34 UTC