Multiagent Systems

The Technology Review article A Smarter Web talks about how the web might evolve from the current HTML-based version to a semantic web. The interesting bits are that the various companies they mention which are making money by analyzing and aggregating the semi-semantically marked up data in private databases. That is, they extract some minimal semantics from the SQL schema of the databases then, after some input from the field experts (I assume), they build RDF schemas to describe the data. They can then build some semi-powerful inferencing engines on top of this semantic knowledge.

This vision of incrementally building ever more sophisticated semantic inferencing capabilities is basically the same thing that Tenenbaum was saying in his AI and Web 2.0 article. It is a compelling vision: not just an end point but a realistic evolutionary path to get there. As we know, great technologies are often abandoned due only to the lock-in power of previous (inferior) technologies, cf. qwerty keyboard, MS windows, NTSC.