Another vote in favour of the NLP take on the Semantic Web

ReadWriteWeb reports on Inform, which is taking a text-crunching, NLP approach to inferring semantic links between their clients' content (and potentially their content and that of other sites). No marked-up content, no RDF, no OWL, no ontologies required. As RWW comments, it would be nice if they at least took their work and used it to create some of the latter to start making some hooks that could tie together the classic vision of the SW with this brute force and NLP-based vision. They're not the only ones doing this, and clearly this vision is running away with the SW project at the moment, although it probably can't (presently) achieve lots of the things that the more formal and explicitly structured approach can. But since the latter is lagging, it would be good to see some bridges being built.