[rdfweb-dev] Tidy up your FOAF week - name and shame

Danny Ayers danny666 at virgilio.it
Sun Jul 11 13:17:32 UTC 2004

Julian Bond wrote:

> Jim Ley <jim.ley at gmail.com> wrote:
>> On Sun, 11 Jul 2004 07:12:11 +0100, Julian Bond
>> <julian_bond at voidstar.com> wrote:
>>> Use of http://xmlns.com/wordnet/1.6/Person instead of foaf:Person
>> This isn't wrong, foaf:Person is just a subClassOf of Wordnet Person,
>> so there's nothing actually wrong with the information, they've just
>> been less specific than they could've been.
> Since we don't have namespace/owl aware parsers that can make the 
> inference, it's useless. It means that the application needs to 
> remember another special case.

You might want to check out the way they're doing 
subclasses/subproperties in the RDFSuite [1] store (the schema-specific 
DB part) : regular RDBMS table for classes, one for properties *and* a 
table for subclass relationships and one for subproperty relationships. 
The inferencing is done by the DB, you just give it some SQL. (They're 
using a bit of object/relational trickery with the instance data table, 
but I'm pretty sure that isn't necessary). I found the link to this in 
the HP paper [2] on mapping to RDBMSs.

I think there's likely a lot more that can be done with existing CMSs 
and such, once it's realised that MySQL *is* an inference engine ;-)

But anyhow, it wouldn't be much good in this particular case, presumably 
the app is interested in foaf:Person but not wn:Person and the 
subclassing is the wrong way around. Hey ho.


[1] The RDFSuite: Managing Voluminous RDF Description Bases (HTML 
<> PDF 
<>), S. Alexaki 
and V. Christophides and G. Karvounarakis and D. Plexousakis and K. 
Tolle, Technical report, ICS-FORTH, Heraklion, Greece, 2000.
[2] http://www.w3.org/2001/sw/Europe/reports/scalable_rdbms_mapping_report/




More information about the foaf-dev mailing list