James Montgomery wrote:
Thank you Thilo for your helpful and thorough comments on Lucene and UIMA
(thread "UIMA and Lucene"). They have helped confirm much of what I
suspected and clarified a number of points.
After further discussions with the engineering company they have indicated
that semantic search may be better suited to their needs (this is an R&D
project conducted between my university and the company, hence why we keep
changing out minds :-). With this in mind we are looking at the following
products. Any comments on their suitability for integration with UIMA to
produce a semantic search solution would be welcome, as would
suggestions of
other semantic search tools.
- Sesame, for storing the CASes produced by our UIMA implementation
- Gnowsis, for searching the knowledge base in Sesame or some other RDF
database (although perhaps I'm showing my ignorance and it only works with
Sesame? :-)
Thanks,
James.
Hi James,
before reading your note, I had never heard of Gnowsis (or Sesame for
that matter, I haven't had occasion to work with RDF). From looking at
the docs, it seems to be a very interesting application and looks like
it may be well suited for UIMA integration. So I'm afraid I can't be
much help here. From looking at their docs for a bit it seems to me
that your idea is worth pursuing, but that's really all I can say.
I know too little about RDF to be sure, but one thing you may want to
consider is how much free text search capability you need. The span
search I talked about in my previous note is basically text search with
some semantics on top. With RDF, you have much more sophisticated
semantics, but no free text search. Perhaps adding Gnowsis, which uses
Lucene internally, will give you the right amount of each.
Ok, enough IFs and MAYBEs ;-) I hope there is something here that you
didn't know already.
--Thilo
