Geospatial Semantic Web Blog

Tim Berners-Lee was challenged by Peter Norvig, Google Director of Search, at a recent artificial intelligence conference sponsored by the American Association for Artificial Intelligence (AAAI).

Tim’s belief is that in the future “Web developers will use semantic languages in addition to HTML. He stressed the importance of using persistent URIs (Uniform Resource Identifiers) and RDF (Resource Description Framework) for identifying information. Consistent use of these specifications, said Berners-Lee, will allow the Semantic Web to maintain the collaborative nature the World Wide Web was originally intended to have.”

At the end of Tim’s speech, Peter challenges Tim with the question that is it really feasible to get Web developers to use semantic languages, given that many of them can’t really configure web server and can’t write HTML?

This is not the first time I heard similar complains from Peter on the Semantic Web. In his 2005 interview, Peter also argued about why it is difficult to use ontologies on the Web.

My take on this issue is as the follows. I admit that the current Semantic Web technologies (RDF, OWL, OWL-S, microformat etc.) are not perfect. They probably can’t solve the kind of semantic problems that Peter is thinking, but they are the best of what we’ve developed so far. We shouldn’t given up the Semantic Web idea just because there are problems it can’t address. It’s quite possible that few years down the road, we will have new technologies that do better than the existing ones, and make the Semantic Web vision happen. If we stop now, we will never get there.

Nevertheless, I do agree with Peter on thing. When building an intelligent web, we must not forget that most users are “incompetent” — they probably can’t edit ontology, don’t understand classification, and are not comfortable using markup languages. For that reason, I believe that any Semantic Web applications we will develop must be simple and intuitive for the everyday people.