OpenCalais

One for the top-down semantic web: Reuters' new API, OpenCalais, that uses AI, NLP and a mahoosive database of people, places, organisations and events to extract semantics from content submitted to it. You can see more about the API here and read the RWW analysis here. It spits out RDF, actually, demonstrating the overlap between the automated, AI-driven SW that is clearly going to happen, and the formal vision of RDF etc. is going to be a real force.

Fotowoosh breaks cover

IIRC, you can download the necessary code to run Fotowoosh or whatever they originally called it yourself, but who wants to do that? Now (I discover, thanks to OUseful) they have made it available through Facebook. I hope I'm not breaking any rules that I've not read about by embedding here an instance that I've made through that site. Here you go, anyway, a view of the Getty Museum with LA in the background. It's evidently not the perfect sort of photo for the job but you get the idea.

IBM Sears Second Life

http://www.3pointd.com/20070108/ibm-brings-sears-to-second-life-at-ces/

Also http://evolvingtrends.wordpress.com/2007/01/07/designing-a-better-semantic-search-engine/ This is in part it's about improving "semantics" search-side i.e. understanding the query better, and in part it's about feeding better info to crawlers, with Wikipedia v 3 in mind

AI beats SW? http://www.roughtype.com/archives/2007/01/the_wikipedian_1.php