Dear Patrick,
Thanks for the references. I wasn't able to access the Yago-SUMO
page. Perhaps there is another source for it?
The website
http://www.mpi-inf.mpg.de/~gdemelo/yagosumo.html
appears to work for me. You can also read the following paper
Gerard de Melo, Fabian Suchanek and Adam Pease (2008). Integrating YAGO
into the Suggested Upper Merged Ontology
Proceedings of IEEE ICTAI 2008. IEEE Computer Society, Los Alamitos, CA,
USA.
Technical Report version:
http://www.mpi-inf.mpg.de/~gdemelo/papers/yagosumo-tr.pdf
I do have the OWL version of the DBpedia ontology, and that is what
I have been looking at as an indicator of the state of the ontology
work. I don't yet know the relation between Yago and the DBpedia
ontology.
DBpedia's ontology is extremely shallow with just 320 classes based on
infobox types.
Many Wikipedia pages do not have infoboxes and hence the corresponding
instances lack
any genuine class. The advantage of this approach is that it is fairly
accurate.
YAGO, in contrast provides several classes for almost all instances with
a Wikipedia page,
but it is a bit less accurate. The class hierarchy contains many
thousands of classes derived
from WordNet, which of course is suboptimal in some ways from an
ontological perspective.
The YAGO-SUMO project aims at replacing the WordNet upper-level of YAGO
with one based
on SUMO.
I am still in the process of trying to get acquainted with the
DBpedia ontology, and aside from what looks like inconsistencies the
structure of the ontology itself, there is a peculiar usage:
Using the SPARQL query page, types of birds like "Albatross"
(http://dbpedia.org/page/Albatross) are retrieved as instances of
"Bird" (<http://dbpedia.org/ontology/Bird>), i.e. the query:
{ ?e <http://www.w3.org/1999/02/22-rdf-syntax-ns#type>
<http://dbpedia.org/ontology/Bird>}
returns a list of types of birds. I would expect these entries to be
**subclasses** of Bird, rather than **instances** (the usual
interpretation of the rdf:type relation). Is this an intentional
variant usage of the notion of "type"?
There have been approaches published on how to recognize Wikipedia pages
that
describe classes rather than individual instances. So far, however, such
work has
not made its way into DBpedia, which treats every page as describing an
instance.
So learning whether there are existing applications that depend on
the current structure of the DBpedia ontology is one of the issues of
primary concern to me in this regard.
People frequently use DBpedia and YAGO as ontologically enhanced databases
that allow for join queries of the sort "Find me all companies founded
in the Bay
Area in the 1960s". The ontology is used to make sure you match all
relevant subclasses of COMPANY. The YAGO2 demo paper has a screenshot [1]
showing how you can visualize the results.
Best regards,
Gerard
[1] http://suchanek.name/work/publications/www2011demo.pdf
--
Gerard de Melo [dem...@icsi.berkeley.edu]
http://www.icsi.berkeley.edu/~demelo/
--
Gerard de Melo [dem...@icsi.berkeley.edu]
http://www.icsi.berkeley.edu/~demelo/
------------------------------------------------------------------------------
Ridiculously easy VDI. With Citrix VDI-in-a-Box, you don't need a complex
infrastructure or vast IT resources to deliver seamless, secure access to
virtual desktops. With this all-in-one solution, easily deploy virtual
desktops for less than the cost of PCs and save 60% on VDI infrastructure
costs. Try it free! http://p.sf.net/sfu/Citrix-VDIinabox
_______________________________________________
Dbpedia-discussion mailing list
Dbpedia-discussion@lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/dbpedia-discussion