- From: Tim Finin <finin@cs.umbc.edu>
- Date: Tue, 11 Aug 2009 07:51:41 -0400
- To: Kavitha Srinivas <ksrinivs@gmail.com>
- CC: Anja Jentzsch <anja@anjeve.de>, Kingsley Idehen <kidehen@openlinksw.com>, "public-lod@w3.org" <public-lod@w3.org>, "dbpedia-discussion@lists.sourceforge.net" <dbpedia-discussion@lists.sourceforge.net>, Chris Bizer <chris@bizer.de>
Kavitha Srinivas wrote: > Actually I wonder if it makes sense to annotate the relation with the > actual conditional probability as suggested by Tim. We found the > probabilities quite useful in different applications. If we go this > route we could also send you the set of types with very low > probabilities -- which is very useful if you want to know for instance > that an instance is almost never both a Car and a Person. Publishing the conditional probabilities available in some form (not necessarily RDF) is a great idea. People could use this data in many ways, I think. It's less clear to me how this might be integrated into DBpedia or the LOD cloud. But having the data available should facilitate experimentation with how to best use it.
Received on Tuesday, 11 August 2009 11:52:37 UTC