Hi Chris, Most of the semantic annotation code is now a bit old (3-4 yrs), and was a prototype. Prior to this version we actually had an approach based on a SPARQL implementation, however, it was decided that Kepler should adopt an "object manager" approach, which stores actors and actor metadata (including semantic annotations of actors) using MoML and within a relational (in-memory) db. So, SPARQL wouldn't work for this, unless we went to an RDF approach for representing annotations, metadata, etc. Another problem with SPARQL (I think this is what you meant) at least from my experience is that it is more focussed on querying resources and their properties, and less useful for querying over class definitions. The latter requires a reasoner (like Pellet, Fact++, etc.). Finding actors based on their semantic types is all about matching class definitions, and not about searching for class instances.
Anyway, we are currently in the process of revamping the annotation interfaces to make it easier and hopefully more obvious to folks that actors (and their ports) can be tagged (e.g., a draft plan for this is here: https://kepler-project.org/developers/incubation/GUI/kepler-tagging). So far, zero Kepler actors have their ports annotated. But perhaps with some UI changes more actors will be annotated, etc. We are also making changes so that basic types of ontology creation and editing are exposed and available from within Kepler. Again, this might make the annotation capabilities more usable. If you would like to help with any of this (designing and/or implement ui's, designing apis, suggesting features, architecture, technology's, etc.) that would be awesome! Shawn On Wed, Jul 1, 2009 at 10:18 AM, Chris Weed<chrisweed at gmail.com> wrote: > Hi, > I have a few questions about the semantic annotation functionality. > I noticed in the semantic search browser it displays the properties, > but I wasn't sure if it used that for doing searches. > > Also, I was looking through the code for performing the semantic actor > suggestion, > and it looks like it uses a hand-coded matching algorithm instead of a SQARQL > based inference engine. Would it make sense to update this functionality > which would allow more sophisticated searches with properties. > > Chris > _______________________________________________ > Kepler-users mailing list > Kepler-users at kepler-project.org > http://mercury.nceas.ucsb.edu/kepler/mailman/listinfo/kepler-users >
