Re: [Entailment] D-entailment (and upwards) issue?

This sounds very reasonable. 

I think it indeed makes sense to require consistency. Ie, if an endpoint implements D-entailment, then either literals in _both_ the graph and the query should be canonicalized (including literals appearing in a filter), or none of the two.

Reading the last paragraph of 4.1 I have the impression that this is the intention (it talks about 'parsing process' which could be seen as applying both on the graph and the patterns), but making this explicit would be good.

Ivan


On May 11, 2010, at 08:53 , Axel Polleres wrote:

> When thinking about what we called "D^-"-entailment at the f2f, I realised the following possible issue for all 
> datatype aware entailment regimes (D-entailment, OWL, RIF, ...):
> 
> Entailment regimes in general are defined only in terms of BGP matching, but nothing else in the algebra, particularly FILTER evaluation
> is independent from the entailment regime. FILTER evaluation is specified solely in the query document at this point.
> 
> However, I am afraid this might lead to unexpected behaviors... take the following example:
> 
> G:  :s :p "1"^^xs:integer
> 
> Query1: 
> 
> SELECT * WHERE {?S ?P "1.00"^^xs:decimal }
> 
> Query2: 
> 
> SELECT * WHERE {?S ?P ?O FILTER(?O = "1.00"^^xs:decimal) }
> 
> 
> Since constants in FILTERs are not affected by canonicalisation, D-Entailment, would only give an answer to Query1, right?
> I find this quite awkard since it'd mean that whenever I pose a query to a datatype aware store, I'd still need to know the 
> internal representation of typed constants when it comes to FILTERs... is this observation correct?
> 
> However, a system that applies canonicalisation of typed constants not only in the graphs but including those in the query, 
> could easily cope with that. I ask myself, whether we couldn't in the Section on canonical representations (4.1) also mention 
> the option that not only the graphs, but also the queries could (or even should?) be canonicalised?
> 
> Any opinions?
> 
> Axel
> 
> p.s.: think I also found a small typo in the section 4.1 .... 
> 
> "If the parsing process involves canonicalization, then the obtained graph will contain just two nodes (one for ex:s and one for the 
>            data value 100.0) connected by one edge (for ex:s)"
> 
> should be:
> 
> "If the parsing process involves canonicalization, then the obtained graph will contain just two nodes (one for ex:s and one for the 
>            data value 100.0) connected by one edge (ex:p)"
> 
> 
> 


----
Ivan Herman, W3C Semantic Web Activity Lead
Home: http://www.w3.org/People/Ivan/
mobile: +31-641044153
PGP Key: http://www.ivan-herman.net/pgpkey.html
FOAF: http://www.ivan-herman.net/foaf.rdf