Re: weird example in RDF-MT document...

Graham Klyne wrote:
> Axel Polleres wrote:
> 
>>Hi all,
>>
>>I was studying the RDF semantics document once again in some detail and
>>it looked to me I found a bug in an example in the end of section 3.3.1.
>>Since I doubt that, I was asking myself whether somebody here can help
>>me to get the knot out of my head....
>>
>>
>>   http://www.w3.org/TR/rdf-mt/#Reif
>>
>>
>>
>>In the end of that section, it is stated that
>>
>>"For example,
>>
>>_:xxx rdf:type rdf:Statement .
>>_:xxx rdf:subject <ex:subject> .
>>_:xxx rdf:predicate <ex:predicate> .
>>_:xxx rdf:object <ex:object> .
> 
> 
>>_:yyy rdf:type rdf:Statement .
>>_:yyy rdf:subject <ex:subject> .
>>_:yyy rdf:predicate <ex:predicate> .
>>_:yyy rdf:object <ex:object> .
> 
> 
>>_:xxx <ex:property> <ex:foo> .
>>
>>does not entail
>>
>>_:yyy <ex:property> <ex:foo> ."
>>
>>
>> This is at the very least strange for me...  and I think simply wrong.
> 
> 
> (From my distant recollection of working group discussion on this point...)
> 
> This example was added to the spec to make a specific point about the nature of
> the "reification" vocabulary in RDF -- it is possible to define the semantics to
> work either way and reasonable people may disagree about whether this is the
> right choice.
> 
> In this case, the nodes _:xxx and _:yyy are effectively described as referring
> to specific instances of statements -- it is possible for two logically
> equivalent statements to be made under different circumstances.
> 
> The motivation for this particular choice was that the more compelling
> reification use cases surveyed were related to provenance applications (who said
> what, and when), which required the (weaker) semantics exemplified above.
> (consider that <ex:property> and <ex:foo> correspond roughly to "said by" and
> "John", and that one might also want to describe a logically equivalent
> statement that was "said by" "Jack", maybe made at a different time (i.e. the
> example offered by Alex).

Dear Graham,


I think I have no problems understanding what the example *means* to 
tell me, but what I simply claim is that

the graph G1:

    _:yyy <ex:property> <ex:foo> .

*is* entailed by the graph G0:

  _:xxx rdf:type rdf:Statement .
  _:xxx rdf:subject <ex:subject> .
  _:xxx rdf:predicate <ex:predicate> .
  _:xxx rdf:object <ex:object> .

  _:yyy rdf:type rdf:Statement .
  _:yyy rdf:subject <ex:subject> .
  _:yyy rdf:predicate <ex:predicate> .
  _:yyy rdf:object <ex:object> .

  _:xxx <ex:property> <ex:foo> .

but the example claims the opposite. if you'd write instead that G0 does 
not entail G2

  _:xxx rdf:type rdf:Statement .
  _:xxx rdf:subject <ex:subject> .
  _:xxx rdf:predicate <ex:predicate> .
  _:xxx rdf:object <ex:object> .

  _:yyy rdf:type rdf:Statement .
  _:yyy rdf:subject <ex:subject> .
  _:yyy rdf:predicate <ex:predicate> .
  _:yyy rdf:object <ex:object> .

  _:xxx <ex:property> <ex:foo> .
  _:yyy <ex:property> <ex:foo> .

then I would think it's fine, but the single triple graph G1 *is* entailed.

  This is more or less what I already replied to denny, who made a 
similar argument to yours. I mean to say that the example is at least 
misleading, easy to be misunderstood and should be clarified... 
although, I don't know whether further work is planned in the nearer 
future. Probably you can give more hints here?

best regards,

Axel



> There were examples raised requiring a stronger form of semantics (i.e. that the
> statement given IS entailed), but they were felt to be more marginal.  But
> mainly, for the sake of consistency, the working group had to make a choice, and
> that was it.  The wider problem of solving semantics for all these use cases was
> out of scope for the working group charter (which was to clarify RDF as
> originally designed).  Personally, I think there are more fruitful ways than
> "reification" to address these problems, such as named graphs.
> 
> #g
> 


-- 
Dr. Axel Polleres
email: axel@polleres.net  url: http://www.polleres.net/

Received on Friday, 9 February 2007 01:22:54 UTC