Re: RDF Recommendation Set comments (re agenda for 6th April)

Le 07/04/2011 18:40, Pat Hayes a écrit :
>
> On Apr 7, 2011, at 11:32 AM, Antoine Zimmermann wrote:
>
>> Le 07/04/2011 17:58, William Waites a écrit :
>>> * [2011-04-07 17:40:37 +0200] Antoine Zimmermann<antoine.zimmermann@insa-lyon.fr>   écrit:
>>>
>>> ] :x owl:sameAs "abc" .
>>> ] :x owl:sameAs "xyz" .
>>>
>>> So this is true, but in the case of catching modelling errors it
>>> is unlikely, I suspect, to crop up. Much more likely is to have
>>> a URI or bnode object where this wouldn't be a contradiction...
>>
>> Honestly, I believe this one actually can be inferred in non-trivial ways on the currently deployed Web of Data, with only RDFS and owl:sameAs. Of course, this is probably not going to be written as is by someone who really believes that the string "abc" is equal to the string "xyz".
>>
>>>
>>> ] rdf:type owl:sameAs owl:sameAs .
>>>
>>> Now that's a funny pathology, but I also suspect it would be
>>> uncommon. I'm more interesting in finding errors in data that is
>>> "almost right", not constructing pathological cases.
>>
>> You can hardly implement a reasoner where you have to manually add special treatment of all the "uncommon" cases that lead to contradictions. Usually, contradictions come from trivial and absurd errors, such as saying that Tim Berners-Lee is a document
>
> While nonsensical, I doubt it that is a logical contradiction to anything. Most errors are not logically inconsistent. I think that relying on consistency checking to find errors is simply a mistake/poor design.

This example was in fact relying on the OWL semantics. From various 
online sources (which I am too lazy today to quote), you can infer that 
TimBL is a foaf:Document and a foaf:Organization (and a foaf:Person but 
I don't need this). From the FOAF ontology, you get that 
foaf:Organization owl:disjointWith foaf:Document. Inconsistency!

Anyway, my point was that there are lots of nonsensical things that just 
look like totally improbable when you extract them out of the existing 
Web of Data, and if one says that it is unexpected to see these 
improbable things expressed on the Web, one may have quite a lot of 
undesired results. In fact, I expect that the equality of distinct 
strings can actually be inferable (with RDFS and owl:sameAs alone) from 
the combination of various badly designed data.

This, again, is not an argument against owl:sameAs, it's just an answer 
to William who said that my example is just a pathological case that is 
not really useful to consider.

>
> Pat Hayes
>
>> or that the resource identified by http://www.w3.org/ is equivalent to the character string "http://www.w3.org/". These things are not uncommon at all in real data.
>>
>>
>>>
>>> Cheers,
>>> -w
>>
>>
>> --
>> Antoine Zimmermann
>> Researcher at:
>> Laboratoire d'InfoRmatique en Image et Systèmes d'information
>> Database Group
>> 7 Avenue Jean Capelle
>> 69621 Villeurbanne Cedex
>> France
>> Tel: +33(0)4 72 43 61 74 - Fax: +33(0)4 72 43 87 13
>> Lecturer at:
>> Institut National des Sciences Appliquées de Lyon
>> 20 Avenue Albert Einstein
>> 69621 Villeurbanne Cedex
>> France
>> antoine.zimmermann@insa-lyon.fr
>> http://zimmer.aprilfoolsreview.com/
>>
>>
>
> ------------------------------------------------------------
> IHMC                                     (850)434 8903 or (650)494 3973
> 40 South Alcaniz St.           (850)202 4416   office
> Pensacola                            (850)202 4440   fax
> FL 32502                              (850)291 0667   mobile
> phayesAT-SIGNihmc.us       http://www.ihmc.us/users/phayes
>
>
>
>
>

Received on Thursday, 7 April 2011 20:07:14 UTC