I agree: semantics on the web are alive and well in works such as schema.org. Also wolfram alpha remains very interesting, although I am less convinced of the sustainability of their business model.

RDF and triple stores are still research projects. Without production ready solutions to authentication, provenance, access control, etc these systems are not ready for the big time yet. And the astronaut-grade musings on inference, OWL, description logics etc will mean that these technologies will not gain easy acceptance amongst the majority of developers.

Schema.org in the meantime will provide an easily implemented route to semantics which plays well with both Google and Bing.