In our last episode, I was trying to construct a list of countries
that are 'going concerns'; that is, not Austria-Hungary. I decided a
reasonable approximation could be had by creating a 1-1 mapping between
dbpedia and country codes. It's easy to get iso country codes out of
Freebase (one MQL query returns them all,) so then it's a simple matter
of using the Fb <-> Dbpedia mappings to send them back to freebase.
Wrong.
After I took out redirects, I found that this procedure mapped ISO
country codes to multiple pages in dbpedia for 90 countries; if these
were all obscure places, that would be one thing, but fbase:/en/japan
mapped to a large number of topics in freebase, even after removing
redirects... Here's some raw triples from the NT dump just so you get
the idea:
<http://dbpedia.org/resource/Area_11>
<http://www.w3.org/2002/07/owl#sameAs>
<http://rdf.freebase.com/ns/guid.9202a8c04000641f800000000001fc6c> .
<http://dbpedia.org/resource/Fauxpan>
<http://www.w3.org/2002/07/owl#sameAs>
<http://rdf.freebase.com/ns/guid.9202a8c04000641f800000000001fc6c> .
<http://dbpedia.org/resource/Japan_%28country%29>
<http://www.w3.org/2002/07/owl#sameAs>
<http://rdf.freebase.com/ns/guid.9202a8c04000641f800000000001fc6c> .
<http://dbpedia.org/resource/Japanese_financial_sector>
<http://www.w3.org/2002/07/owl#sameAs>
<http://rdf.freebase.com/ns/guid.9202a8c04000641f800000000001fc6c> .
<http://dbpedia.org/resource/Administrative_divisions_of_Japan>
<http://www.w3.org/2002/07/owl#sameAs>
<http://rdf.freebase.com/ns/guid.9202a8c04000641f800000000001fc6c> .
There are a total of 44 of these in the dump. A lot of these are
redirects to dbpedia:Japan (easy to zap) and they all make sense as
being aspects or components of 'Japan', even in some of them are a
little batty, like "Area 11".
For me, there's the immediate problem that, if I want to key my
items to dbpedia, I need to establish (more or less)
<dbpedia:Japan> <owl:sameAs> <fbase:/en/japan>
and that's it. So far as end users are concerned, there's got to
be some single entity that corresponds to a country, and that entity
has to be correct.
Now, for me, the practical answer to use fbase:/en/japan as my key.
Those of you who are using OWL reasoning, however, had better
note that adding the freebase->dbpedia mappings to your copy of dbpedia
will considerably change the meaning of your database.
------------------------------------------------------------------------------
_______________________________________________
Dbpedia-discussion mailing list
[email protected]
https://lists.sourceforge.net/lists/listinfo/dbpedia-discussion
