Before I read Bjørn Sandvik's post about semantic geospatial services, I didn't know there was an effort to build bridges from WS-* to the Semantic Web. More significantly, his post is a useful introduction to semantic web issues to GIS users who are more familiar with the concept of services than the concept of Web resources. The metadata in OGC service capability documents (abstract, keywords, etc) are inadequate for the purpose of machine reasoning, and application of RDF and OWL could certainly make it more clear what a service is about.
There is a different effort already underway to semantically link open datasets called "Linked Data" [wikipedia, http://linkeddata.org] or "Linking Open Data" [W3C]. This service-free, or at least, service-agnostic, approach is the one I find more compelling for Pleiades and Concordia. The diagram below (from Richard Cyganiak) shows how geographic data courtesy of GeoNames is right at the heart of the Linked data project.
This graph of data has grown by leaps and bounds since we pitched Concordia. We didn't initially propose to join the Linked Data project, but I'd really like to see Ancient World datasets link themselves together in this way.