- From: Sandro Hawke <sandro@w3.org>
- Date: Fri, 05 Sep 2003 16:33:38 -0400
- To: public-sw-meaning@w3.org
I work for W3C Semantic Web Advanced Development (SWAD) at MIT, and participate in WebOnt and the DAML Joint Committee (mostly for Rules work). I've been thinking about how URIs can or should work in RDF since about Dec 2000, when I was hired by W3C. Before that, I called HTTP URIs "URLs" and thought of them as communication end-points, not global object identifiers. My first reaction to seeing RDF's use of URIs was to propose the "tann:" URI scheme (since renamed "tag:" and slowly moving through the IETF) http://www.w3.org/2001/02/tann/ to provide URIs which are free from all this confusion about web operations. Tann/Tag: URIs, by definition, have no 'dereference' mechanism. I have since realized that being able to do web operations on the URIs occurring in RDF is actually quite useful, so I don't generally recommend the use of tag: URIs. I mention them now because I still think they provide a useful reference point in this area; why would someone prefer http: URIs over tag: URIs without clarifying how dereferencing the http: URI relates to its meaning? I've jumped into the water a few times since then, and usually came out confused and frustrated. But every time I try to use RDF, I find I need URIs, and I can see problems with just about every way they could be used. Like Dan Connolly, I have generally tried to gather my conclusions into the ESW Wiki, starting such pages as http://esw.w3.org/topic/GoodURIs http://esw.w3.org/topic/HashVsSlash which can serve as something of a position paper for me. In particular, I seem to be the most vocal proponent of using HTTP redirects: http://esw.w3.org/topic/SlashRedirection because it seems to solve the most problems (of the ones I know/care about, at least!). I called the Budapest BOF in the hope of getting the ball rolling again after the SWCG had gently set it on the ground. After the BOF, I, um, bugged the SWCG until it handed the issue off to the TAG. Now it's come back and I'm willing to co-chair the group, but I'm not attached to the role. I am motivated to see the group succeed, however. Meanwhile, my background is in network software development, including protocol design. These days I code happily in Python, only occasionally missing C and C++. I don't seem to miss Perl or Java (or languages from the 70s and 80s). I'm also fairly comfortable with "programming" in Prolog and First Order Logic, which I usually run with XSB or SWI, and Otter. (I don't care for Prolog as a programming language, but it can be a nice tool.) I recently read _Naming and Necessity_ (Saul A. Kripke) and the relevant parts (like "Causal Theory of Names") of _The Cambridge Dictionary of Philosophy_ (Robert Audi, Ed). I think it's helped my understanding here. I had been thinking along the Russellian lines of terms standing for definite descriptions; Kripke has shredded that idea for me. Also, I'll be somewhat less lost if people start quoting Wittgenstein again.... I think "rough consensus and running code" is our best protection when diving into age-old philosophical issues. Use cases, test cases, demonstrable working systems. But I also want it all to make sense; I'm highly sympathetic to Pat Hayes' critiques of RFC 2396. I'm flexible as far as timing and pacing. I think my only fixed-time committments are shared with several others in this group, and I'm okay with trying to do this quickly/intensely or slowly/intermittently. Hm. I'm feeling optimistic about the group, and eager to hear what folks who really want to solve this problem (instead of just making it go away!) have to say about it. -- sandro http://www.w3.org/People/Sandro
Received on Friday, 5 September 2003 16:33:39 UTC