Saturday, June 18, 2005

G8 and Indymedia take over

I'm going to apologise to my academic advisors for this, but I'm going to have to temporarily slacken my academic duties in order to deal with Indymedia and the upcoming G8 protests. For news about the G8, check out the site I help run Scotland Indymedia, and the main site G8 reporting on Indymedia UK.



The G8 protests...and Indymedia..take over Edinburgh, setting up their centre in the Forest Cafe right north of the University!

Thursday, June 09, 2005

W3C Semantic Web Services Workshop

Surprise, surprise - I'm the opening speaker at the W3C Semantic Web Services workshop in Innsbruck, Austria! My presentation (with commentary too!) and paper are online, both a sort of abbreviated and clarified take on my YR-SOC 2005 paper.

My point is rather simple: the Web needs a good unifying framework to connect the ordinary hypertext Web we all know and love, with the rather visionary if not entirely working correctly Web Services and Semantic Web visions. My vision is rather simple The Web can be unified as a universal computation space, and to do this one needs just two things: functions and types for Web-distributed computing. Luckily for us, the Semantic Web ontologies are actually useful for something: specifying the real-world denotations of data in a consistent model-theoretic manner, and so can serve in conjunction with XML Schema data-types as a type system, and Web Services are clearly functions.



Presentation by the ever-enjoyable Bijan Parsia


Everyone seemed to enjoy the presentation from "the guy whose so loud he doesn't need a microphone" (much like Henry Thompson!), although I did get sniped at for two angles: one for not using "process algebra" and instead using pi-calculus (although I believe this was a disguised jump to get FLOWS on board, and my noticeable lack of mention of OWL-S, although I might add that OWL-S has no model-theoretic semantics anyways, although I think Barry Norton just retrofitted one) and for not mentioning non-functional properties. I would respond that non-functional properties could just be included as a constraint on the input. For example, if one wants a Web Service in North Dakota that takes names and gives credit histories, one should just have the Web Service as f(name, (location="NorthDakota"))=>creditHistory. Or one could conceive of the Web Service as a triple. The important thing is that non-functional properties are a misnomer, these properties clearly function to constrain what type of service one invokes. Lastly, the results of the workshop are going my way, and here's me after an enjoyable dinner with the workshop participants, looking over the top of a mountain overlooking the Alps!



Yes, that's the Alps behind me!