- From: Phillip Lord <phillip.lord@newcastle.ac.uk>
- Date: Fri, 30 Apr 2010 09:31:58 +0100
- To: Chris Mungall <cjm@berkeleybop.org>
- Cc: Alan Rector <rector@cs.man.ac.uk>, Owl Dev <public-owl-dev@w3.org>, sonic@tcs.inf.tu-dresden.de
Chris Mungall <cjm@berkeleybop.org> writes: > On Apr 29, 2010, at 4:08 AM, Alan Rector wrote: >> Or there's the least common named subsumer in the inferred >> classification lattice using lattice theoretical tools, but I presume >> that is not what you are asking or you wouldn't be asking it. > > Correct, it would be a solved problem if the result was drawn only > from the set of named classes. Okay, so email overlap. It seems reasonable to me to assume that at the time you want to calculate a semantic similarity, then you have all the three terms that you want -- the two that you wish to compare, and the (unknown, explicitly expressed in the ontology) term that is the LCS. I can see a very strong use case why you might want to allow the query terms to not pre-exist, but why the LCS? What semantic similarity measures were you thinking of anyway? The information content based ones will, I think, require that the LCS pre-exist anyway. Otherwise, you will have to test each LCS against the entire corpus for each query. Phil
Received on Friday, 30 April 2010 08:32:38 UTC