Tuesday, November 11, 2003

Clay Shirky describes the Semantic Web in the course of eviscerating the entire concept. I had sort-of-heard of this idea, but I hadn't really paid any attention, because it showed all the signs of a Wired Special. Wired Specials are flack-friendly "revolutionary concepts" which can only be described in marketing jargon, which don't relate to any actual pre-existing problem or problems, and seem to exist mostly as a totem of enthusiasm and/or fervor. "Push" technology was the classic Wired Special.

As far as I can tell from Shirky's article, the Semantic Web is a data-mining project based on the assumption that you can extract new ideas from syllogistic evaluation of the existing web. I have to wonder if a failure of the Semantic Web suggests by extension the failure of data mining in general? Well, no. What is our HighQ project but a data-mining system for agricultural data? Perhaps it was the abstract scale of the Semantic Web that puts the knife in. But most data-mining systems are inductive in nature - summary results form new hypotheses, and queries are made to product new, more specialized results. Repeat next year to test the work done under the hypotheses of this year. Syllogistic reasoning is a deductive approach to an essentially inductive task.

Via Jeff Jarvis.

No comments: