Re: Restructure definition of Basic Graph Pattern and pattern match (sec 2.4)

Jeen Broekstra wrote:
> SPARQL language editor's draft rev 1.379
> http://www.w3.org/2001/sw/DataAccess/rq23/
> 
> In section 2.4 the terms 'Pattern Solution', 'Query Solution' and
> 'Basic Graph Pattern' are defined,
> 
> The definition of Basic Graph Pattern includes what it means to match
> against a graph. It would be better to create a separate definition
> for this, called 'Graph Pattern Match'. The definition of 'Basic Graph
> Pattern' can then be moved to the start of the section, and in any
> case I think an explicit definition of pattern match, that links graph
> patterns to solutions, is a lot clearer.
> 
> Also, the current definition of a match is imprecise. It uses the
> notion 'entails' without specifying what that means. Is that simple
> entailment, RDF(S) entailment, or an entirely different, more loose
> form of entailment? I request instead using the notion of 'subgraph'
> for defining a match.

It has already gone back to being "subgraph" to avoid entailment.  Simple 
entailment does not add anything (except a lot of blank nodes).  "Entailment" 
does not define what kind.

The match can be a subgraph of an entailed graph, putting the entailment outside 
the SPARQL doc.

> 
> Suggested definitions:
> 
>    Definition: *Basic Graph Pattern*
> 
>    A /basic graph pattern/ is a set of Triple Patterns.
> 
>    Definition: *Graph Pattern Match*
> 
>    A graph pattern /matches/ on a graph G with pattern solution S iff
>    S(GP) is a non-empty subgraph of G.

This is at odds with later definitions.  This defn is only basic patterns.

I have left order as is but 2.4 is split:

2.4 Pattern Solutions
2.5 Basic Graph Patterns

and old 2.5 (example) is part of new 2.5 as well.

> 
> Jeen

	v1.391
	Andy

Received on Monday, 13 June 2005 13:51:27 UTC