RE: Definition of Basic Graph Pattern Matching & Pattern Instance Mapping



> -----Original Message-----
> From: public-rdf-dawg-request@w3.org [mailto:public-rdf-dawg-request@w3.org]
> On Behalf Of Birte Glimm
> Sent: 24 September 2009 15:00
> To: SPARQL Working Group
> Subject: Definition of Basic Graph Pattern Matching & Pattern Instance
> Mapping
> 
...

> First, a pattern instance mapping P is defined as a composition of an
> RDF instance mapping sigma with a solution mapping mu, i.e.,
> P(x)=mu(sigma(x)), see 12.3.1. The domain of an RDF instance mapping
> is, however, the set of blank nodes and its range is the set of
> literals, blank nodes, and IRIs. Now the domain of mu is the set of
> variables, which means we cannot really compose the two. The range of
> the first is disjoint from the domain of the second. What I would
> rather say is that P is a pair (mu, sigma).

I've put this on the errata page. We could use a form of wording in the entailment doc and remove it from there.

While I agree the wording can be improved, I believe the usage "Pattern Instance Mapping" is using the functions as the implied mapping from pattern to pattern.

 Andy

Received on Monday, 5 October 2009 12:02:19 UTC