Re: typed literal equality

Pat Hayes wrote:
> On Apr 24, 2011, at 12:37 PM, Nathan wrote:
> 
>> moving to RDF WG list:
>>
>> Andy Seaborne wrote:
>>> On 24/04/11 14:42, Nathan wrote:
>>>> so are we saying that the following are all different for purposes of
>>>> establishing equality?
>>>>
>>>> "100"^^xsd:double
>>>> "1E2"^^xsd:double
>>>> "1e2"^^xsd:double
>>>> "+100"^^xsd:double
>>>> "+1E2"^^xsd:double
>>>> "+1e2"^^xsd:double
>>> They are all different as RDF terms.
> 
> Because they are: look at them.
> 
>>> They are the same *value*, different ways of writing it.
> 
> This is determined by the datatype mappings themselves, which often map distinct forms to the same value. 
> 
>>>  RDF (simple entailment) does not say anything about literal values.
>>> They are the same under XSD D-entailment (which requires full RDFS)
>> Why?
> 
> Why have several entailments? Good question. This was a basic design decision for RDF, to assign each namespace its own semantic conditions, so that we have simple entailment (just based on the graph syntax itself) , RDF entailment (basically the XMLLiteral stuff), RDFS entailment and RDFS-with-datatype entailment. This last one has to be defined with a parameter, D-entailment, because (of course) it varies depending on what datatypes you want to have recognized. 
> 
> As you maybe can tell, there was a strong sense of keeping things modular so that people could choose the 'pieces' they wanted without being committed to the whole ball of wax. And also (at the time) we didn't really have a good sense of what pieces people would want, so we probably did not carve it at the joints.
> 
> That help?

Certainly does, thanks Pat. It leads me on to many other questions of 
the "could we" variety, but I think it's probably best to leave this 
train of thought here.

Best,

Nathan

Received on Sunday, 24 April 2011 22:40:24 UTC