ISWC 2003, Sanibel Island, Florida. http://iswc2003.semanticweb.org/ Trip report by Jacco. Overall impression: conference is much bigger than previous year, with three tracks the first day and two for the other days (c.f. only one track last year). More than 480 attendants, while 300 was expected (rumor: only 300 proceedings printed ...). Another good thing: 2002 was only about theory and foundations, this year is about 50/50 applications and foundations. ISWC2003 even has a industrial track and a multimedia session. Monday 20 October I gate-crashed the morning session of the scalability workshop, with talks from Jan Wielemaker (SWI-Prolog) and Jeen Broekstra (Sesame). Impressed by speed of Jan's stuff (need to look at his triple20 RDF editor when we are back) and Soton's 3store (http://www.aktors.org/technologies/3store/) used in the AKT demo). Also good to see that most implementations have some notion of context associated with each triple, with Sesame as the exception to the rule (Jeen: you've the source, if you want context ... :-). Worked with Joost on slides and demo in the afternoon. Tue 21 October Multimedia session. First talk was a bit naive approach to image annotation (added later: the speaker actually apologized to Guus after seeing the MIA demo, which Guus and I both thought was a pretty courageous thing to do). Second was typical INA talk on AV-annotation with a hybrid OWL/Semantic Web approach. Next was SCULPTEUR, your typical big European project that just started talk. Next was Joost (see video later), got some people talking with us after the talk and in the breaks over the day. I half missed the FrameNet talk, but it was basically about a WordNet based on frames, so with more ontological richness. Next talk didn't show up. Jim Hendler's after lunch keynote was your typical "a little semantics goes a long way" scruffy talk. Funny to see the old Daghstuhl slide from FrankvH about how the semantic web is different from good old fashioned KR, that Lloyd, and I contributed to a lot. I missed the Ontotext talk (room was full!) but they quote our Jodi paper in a funny way. See proceedings. Next was the award ceremony of the SW Challenge, won by the AKT demo of Nigel's Soton crowd (impressive 24/7 real world data harvesting). Then a security policy talk about Rei (including the deontic stuff you expect, but also speech acts). Next was Jan Wielemaker, but this was basically the same talk as during the workshop. Wed 22 October C-OWL talk (co-authored by FrankvH and Heiner) about a contextualized version of OWL seemed, informally, like a Semantic Web equivalent of XML-namespaces (Frank could live with this view). I thought it was so much better and so much more web-like than the current OWL that I wondered about why OWL wasn't done like this from the start. Frank said the DL people and the lack of reasoning support were the main reasons. Magpie talk by John Domingue from KMI was really impressive, while scientifically they may do not even that more than COHSE did. But is was well worked-out with a good talk and good demos. Missed most of Deborah's SemWeb explanations talk writing this trip report, but it seems like a good article to cite if we need to defend the process related metadata we want to add to Cuypers, with good lists of why you want to do this and what type of metadata you need to explain your answer. RVL talk about an RDF View Language was a bit disappointing, to much a literal copy of RDBMS views to RDF. And maybe I missed it, but the key insight that the result of an RDF query should be another RDF graph and not a table of variable bindings like RQL was already made by Sesame's construct queries. Still convinced that views are useful, but probably, like query languages, only if standardized and applicable across RDF implementations (note adder later: this paper won the best paper award, but nobody I talked to agreed with this). Brodie's key note was all about industrial strength web services, with many 100 billion and trillion dollar slides... Basic message: web services is, like CORBA and java, not the next silver bullet. But it is next major step. But I lost my attention after the first trillion dollar slide. Being not overly impressed by the Web Services talks I heard in the past, I decided to go to the other session. This was a mistake, since the talks were quite boring and we missed Marta's talk - I hadn't realized she was in the services session :-( Anyway, Joost and I did some serious Cuypers debugging while sitting in the back. Thu, 23 October. Haystack talk was a typical MIT talk: slick without any technical content. They could have given this talk at a high school, but I thought it was almost embarrassing on a technical conference like this. Anyway, Haystack is this suite of desktop applications, including mail, calendar, address books, etc., all integrated by using only RDF data formats. But how they align the RDF schema's etc remains a mystery. Joost and I skipped the rest of the Thursday morning talks to go biking around the island, which was pretty cool. Saw a big 'gator (http://media.cwi.nl:8080/geurts/pics/florida/) in J. Ding Darling national park (for the SW insiders, the name was almost too good to be true). Just back in time for Tim Berners-Lee's talk, which was quite good. The first half of the talk was about how to spread the word of the SW while being aware of the risks. Second half of the talk was a good road map overview, with the need for SW query languages, rules and integration with WS (Web Services). TBL's talk ended the conference. Like any conference, the useful bit was not necessarily in the talks (while I really enjoyed several of them). Chatted to a lot of A'dam people that I somehow don't chat enough with when at home (Frank, Guus, Jan). All in all, a very fun and useful conference. We better start thinking about the 2004 paper (ISWC in Hiroshima, Japan): 2005 will be in Galway, Ireland...