I'm loading dbpedia ontology type assignments from 3.5.1 into my system 
and noticing a significant number of subjects that don't have labels...  
Here are just a few I saw going by:

<http://dbpedia.org/resource/Fiat_1300/1500__1295_cc_OHV_I4> 
<http://www.w3.org/1999/02/22-rdf-syntax-ns#type> 
<http://dbpedia.org/ontology/AutomobileEngine> .
<http://dbpedia.org/resource/Fiat_1300/1500__1295_cc_OHV_I4> 
<http://www.w3.org/1999/02/22-rdf-syntax-ns#type> 
<http://dbpedia.org/ontology/Device> .
<http://dbpedia.org/resource/Fiat_1300/1500__1481_cc_OHV_I4> 
<http://www.w3.org/1999/02/22-rdf-syntax-ns#type> 
<http://dbpedia.org/ontology/AutomobileEngine> .
<http://dbpedia.org/resource/Fiat_1300/1500__1481_cc_OHV_I4> 
<http://www.w3.org/1999/02/22-rdf-syntax-ns#type> 
<http://dbpedia.org/ontology/Device> .
<http://dbpedia.org/resource/Adventuress_%28schooner%29__ADVENTURESS__Schooner_Yacht>
 
<http://www.w3.org/1999/02/22-rdf-syntax-ns#type> 
<http://dbpedia.org/ontology/Place> .
<http://dbpedia.org/resource/Adventuress_%28schooner%29__ADVENTURESS__Schooner_Yacht>
 
<http://www.w3.org/1999/02/22-rdf-syntax-ns#type> 
<http://dbpedia.org/ontology/HistoricPlace> .
<http://dbpedia.org/resource/Jos%C3%A9_Antonio_Delgado__Mountaineer> 
<http://www.w3.org/1999/02/22-rdf-syntax-ns#type> 
<http://dbpedia.org/ontology/PersonFunction> .
<http://dbpedia.org/resource/Imogen_Lloyd_Webber__Author> 
<http://www.w3.org/1999/02/22-rdf-syntax-ns#type> 
<http://dbpedia.org/ontology/PersonFunction> .
<http://dbpedia.org/resource/Gary_Mule_Deer__Actor_Comedian_musician> 
<http://www.w3.org/1999/02/22-rdf-syntax-ns#type> 
<http://dbpedia.org/ontology/PersonFunction> .

Some interesting stuff seems to be going on here,  and I'd like to see 
it better documented

For a long time there's been the problem that some named entities have 
their own DBpedia resource and others don't because of the accidental 
nature of how things work on Wikipedia.  Here's an example of two 
fictional characters that play an isomorphic role in two very similar works:

http://en.wikipedia.org/wiki/Sailor_Mars
http://en.wikipedia.org/wiki/Yes!_PreCure_5#Cure_Rouge

The perfect "generic database" would treat these two entities in the 
same way...  And it looks like dbpedia is starting to address this.

http://en.wikipedia.org/wiki/Fiat_1300

has two engines listed in the Infobox,  and it appears the extractor is 
parsing the "name" of the engine to extract some nice facts:

http://dbpedia.org/page/Fiat_1300/1500__1295_cc_OHV_I4

On the other hand,  this one looks like

http://dbpedia.org/resource/Adventuress_%28schooner%29__ADVENTURESS__Schooner_Yacht

doesn't have such clear value...  It seems to be identifying a "facet" 
of the boat,  that is,  its career as a part of the U.S. National 
Register of Historic Places.  The value of this isn't so clear to me,  
but perhaps there is some sense to it.  The oddly named 
dbpedia-owl:added field applies to this career,  so maybe this makes sense.

The two PersonFunctions up there look specious,   particularly because I 
don't see

dbpedia-owl:Writer
dbpedia-owl:Actor
dbpedia-owl:Comedian
dbpedia-owl:MusicalArtist

anywhere near the Person or PersonFunction instances.  That said,  I do 
see some places where "PersonFunction" has promising roles to play:  the 
functionEndData and functionEndYear properties are obviously useful.

There's an issue of how these new "synthetic" identifiers (that don't 
correspond 1-1 to wikipedia pages) map to external resources.  
"PersonFunction" is similar (but probably not equivalent)  to the 
"Employment Tenure" CVC in Freebase.  Certainly some of these new 
synthetic identifiers correspond to resources in Freebase as well,  
since Freebase sometimes splits wikipedia topics into multiple entities.

Note also looking at

http://mappings.dbpedia.org/server/ontology/classes/PersonFunction

I see a lot of properties that just don't make any sense at all,  such a 
"wavelength",  "sourceConfluenceMountain",  "shipDraft".  I don't know 
if this some artifact of how the whole system works (looks like these 
properties all have a domain of "owl:Thing") but it does shock my 
sensibilities.  (From a strictly owl wavepoint,  I guess that something 
that has a "wavelength" is a thing,  but by that standard,  anything 
that has any property at all is a "Thing")



------------------------------------------------------------------------------
_______________________________________________
Dbpedia-discussion mailing list
Dbpedia-discussion@lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/dbpedia-discussion

Reply via email to