Richard,

On 16 May 2008, at 14:16, Richard Light wrote:
>>> That is, that dates of birth and death
>>> appear as different properties:
>>>
>>> p:birthplace
>>> p:birthPlace
>>> p:cityofbirth
>>> etc.
>>>
>>> I can see how this diversity arises, due to the harvesting approach
>>> used. However, (and I'm probably showing my ignorance here) couldn't
>>> owl:sameAs statements be added to the database to indicate that  
>>> these
>>> are all actually the "same" property?
>>>
>>> p:birthplace owl:sameAs p:birthPlace
>>> etc.
>>
>> Yes, this could be done. The problem is that Wikipedia is huge, and  
>> there are tens of thousands of properties covering all sorts of  
>> domain. Hence we would need either an automated approach to find  
>> those duplicated properties, or lots of volunteers who go through  
>> the dataset and find them manually.
>
> Presumably it's possible to put in a SPARQL query to find all the  
> properties that have been used?  All the properties that relate to  
> people?
>
> I'm coming into this from a historical perspective (specifically  
> museums), and my interest would be in sorting out the properties  
> which relate to:
>
> - people
> - places
> - events
> - dates
> - [museum] objects
>
> Would this be a manageable subset?  If so, I'm willing to help.  It  
> would be interesting, additionally, to map these properties to those  
> in the CIDOC Conceptual Reference Model.

You can get an idea of the properties used for a particular kind of  
resource by following these steps:

1. Find the class(es) used in DBpedia of that kind of thing. For  
example, if you look at a couple of persons, you will find that they  
usually have the classes yago:Person100007846 and foaf:Person. Classes  
in DBpedia are quite inconsistent, so this is not an easy task.

2. Use a SPARQL query like this to find the properties used for any  
particular class:

SELECT DISTINCT ?p
WHERE {
?s ?p ?o .
?s rdf:type <http://dbpedia.org/class/yago/Person100007846> .
}

You can run the query at http://dbpedia.org/snorql .

3. Use a SPARQL query like this to find out how often a particular  
property is used (to see if it's worth spending time on it):

SELECT COUNT(*) WHERE { ?s dbpedia2:dateOfBirth ?o . }

4. Find example triples that use a particular property (to see *how*  
it's used, and if it's indeed interchangeable with another property):

SELECT ?s ?o WHERE { ?s dbpedia2:dateOfBirth ?o . }

The end result of this process could be triples like this:

dbpedia2:dateOfBirth owl:equivalentProperty dbpedia2:birthdate .

We could load this into the SPARQL endpoint then.

Best,
Richard




>
>
> Richard
> -- 
> Richard Light
> XML/XSLT and Museum Information Consultancy
> [EMAIL PROTECTED]
>

-------------------------------------------------------------------------
This SF.net email is sponsored by: Microsoft 
Defy all challenges. Microsoft(R) Visual Studio 2008. 
http://clk.atdmt.com/MRT/go/vse0120000070mrt/direct/01/
_______________________________________________
Dbpedia-discussion mailing list
Dbpedia-discussion@lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/dbpedia-discussion

Reply via email to