W3C home > Mailing lists > Public > semantic-web@w3.org > December 2010

Re: The Two Dogs Problem

From: Martin Hepp <martin.hepp@ebusiness-unibw.org>
Date: Fri, 17 Dec 2010 10:59:46 +0100
Cc: Melvin Carvalho <melvincarvalho@gmail.com>, Semantic Web <semantic-web@w3.org>
Message-Id: <5C935016-9F69-4667-8D6C-5B2B5D913F8E@ebusiness-unibw.org>
To: Dave Reynolds <dave.e.reynolds@gmail.com>
Hi Melvin,
my two cents ;-) :

1. For owns, there is already a property read for use: http://purl.org/goodrelations/v1#owns

2. As Dave indicates, the irritation comes from the difference between  
"a" in English and in Turtle. In English, it's a quantifier, in Turtle  
it means "instance of".

3. Using DBPedia URIs as classes is problematic, because they usually  
represent topic instances. But Yago or Cyc will provide nice classes  
for dogs.

3. BNodes and resources identified by a full IRI both leave open  
whether both dogs are the same, so

    foo:Fix a  vocab:Dog .
    foo:Foxy a  vocab:Dog .
    foo:Alice gr:owns foo:Fix, foo:Foxy .

does not yet make clear that these are two dogs; it could also just be  
two names / two identifiers for the same entity.

For stating that, you would have to use owl:differentFrom :

    foo:Fix owl:differentFrom foo:Foxy .

Then you could infer that she owns at least two dogs.

Best
Martin



On 17.12.2010, at 10:30, Dave Reynolds wrote:

> On Fri, 2010-12-17 at 09:55 +0100, Melvin Carvalho wrote:
>> Apologies if this has come up before.  I was wondering the best way  
>> to
>> model the following:
>>
>> Alice owns a dog.  We choose to model it as follows.
>>
>> :Alice :owns a dbpedia:dog
>
> That's an RDF statement, maybe you meant:
>
>   :Alice :owns [a dbpedia:dog] .
>
> I.e. there is something, which is a dog and which Alice owns.
>
>>
>> All is going well.
>>
>> Then Alice gets given a second dog.
>>
>> If we write:
>>
>> :Alice :owns a dbpedia:dog
>> :Alice :owns a dbpedia:dog
>
>   :Alice :owns [a dbpedia:dog] .
>   :Alice :owns [a dbpedia:dog] .
>
> is a reasonable graph, it is not lean since both those bNodes might
> correspond to the same dog but it is a graph with 4 statements not a
> graph with two statements.
>
> To convey that those are definitely different dogs then you can add an
> owl:differentFrom link:
>
>   :Alice owns _:1 .
>   _:1 a dbpedia:dog .
>   :Alice owns _:2 .
>   _:2 a dbpedia:dog .
>   _:2 owl:differentFrom _:1 .
>
> Dave
>
>
>
Received on Friday, 17 December 2010 10:00:18 GMT

This archive was generated by hypermail 2.3.1 : Tuesday, 26 March 2013 21:45:40 GMT