It depends on what do you want to do.
The triples with #10 are compiled using some heuristics and data from
DBpedia in 16 languages, while #9 to #6 are automatically calculated
using the previous triples as training. Usually #9 is very precise (more
than 90%).
I ask you not to edit the type, as I prefer that the Airpedia label
lasts in the triples.
When the Italian DBpedia will release the new dataset, I'll send an
e-mail to the community explaining how we decide to use the w3 type
leaving a link to Airpedia.
Feel free to ask more information.
Best,
Alessio
Il 19/06/13 13:38, kasun perera ha scritto:
So these entity types are a new set separated in to languages.
I can see these triples are following format
<http://be.dbpedia.org/resource/Паўночная_Ірландыя
<http://be.dbpedia.org/resource/%D0%9F%D0%B0%D1%9E%D0%BD%D0%BE%D1%87%D0%BD%D0%B0%D1%8F_%D0%86%D1%80%D0%BB%D0%B0%D0%BD%D0%B4%D1%8B%D1%8F>>
<http://airpedia.org/ontology/type_with_conf#10>
<http://dbpedia.org/ontology/Place>
So i'm going to import this in to dbpedia as follows; can someone
confirm whether i'm right?
<http://be.dbpedia.org/resource/Паўночная_Ірландыя
<http://be.dbpedia.org/resource/%D0%9F%D0%B0%D1%9E%D0%BD%D0%BE%D1%87%D0%BD%D0%B0%D1%8F_%D0%86%D1%80%D0%BB%D0%B0%D0%BD%D0%B4%D1%8B%D1%8F>> <http://www.w3.org/1999/02/22-rdf-syntax-ns#type> <http://dbpedia.org/ontology/Place>
type_with_conf#10 means accuracy around 100% right? So is it a good
idea to import only entries having config#10 or just import all
without worrying about it?
On Tue, Jun 18, 2013 at 9:19 PM, Alessio Palmero Aprosio
<[email protected] <mailto:[email protected]>> wrote:
Dear Kasun,
on the web page
http://www.airpedia.org/download/dbpedia-entity-types-in-31-languages/
you can find the classes in 31 languages.
The type is set to http://airpedia.org/ontology/type_with_conf#X ,
so you can import the triples into DBpedia without conflicts.
In the next days, the Italian DBpedia team will release a new
extraction (based on DBpedia 3.8) with these triple in it. The
format in this case is slightly different: I'll write about it in
the next e-mail.
Let me know if you have troubles.
Best,
Alessio
Il 18/06/13 03:10, kasun perera ha scritto:
I started working on integrating Airpedia types to Dbpedia-links.
As Marco mentioned previously, these are the 3 data-sets that I
can use and I need to analyze what is the best data set that to
be integrated in to the Dbpedia.
<http://airpedia.org/extraction/10-all>
extracted using classifier with all kernels and the bottom-to-top
strategy.
<http://airpedia.org/extraction/10-all-top>
extracted using classifier with all kernels and the top-to-bottom
strategy
<http://airpedia.org/extraction/10-tpl>
extracted using classifier with the sole template kernel and the
bottom-to-top strategy
In Alessio et. al. ESWC paper, I can see it has already done some
tests to measure precision and recall of these classifications.
Can I use those results to select best Airpedia data-set for
dbpedia? or do I need to test with another data set(the ones
Aleksander mentioned) to measure precision and recall?
Thanks
--
Regards
Kasun Perera
--
Regards
Kasun Perera
------------------------------------------------------------------------------
This SF.net email is sponsored by Windows:
Build for Windows Store.
http://p.sf.net/sfu/windows-dev2dev
_______________________________________________
Dbpedia-developers mailing list
[email protected]
https://lists.sourceforge.net/lists/listinfo/dbpedia-developers
