Finally, had to give my talk about the gnowsis adapter framework. Ansgar reviewed my slides in the morning. He was content with them (luckily he didn’t see the version of wednesday 6pm). We went to the Radisson, came late, missed a part of the keynote which was about UIMA, the IBM NLP framework that has been buzzing around for a longer time anyway.
ok, the final version of the slides is here.
10:30 – my talk started. I knew I had 20 minutes before our Aperture release, so everything decent. Only problem was that my Acer laptop fucked up and couldn’t display on the beamer anymore. Damn these laptops, the always brake when you need them most. Pressing fn+screen didn’t do anything, the laptop just refused to work. It was 10:28 and I knew there was no time for reboot now. I experienced this behaviour before and knew a hack-remedy: pressing the fn+screen combination hysterically often, like in a weird C64 game. After 10 seconds of hammering, it did the trick, the slides were on the beamer. But I couldn’t switch to presentation modus. I planed to use the powerpoint presentation modus to know what slides is next and read my notes. Without my notes, I was pretty blind then, so I decided to go in attack mode and switched to my “presenting by walking around and big gestures” mode, where I try to fill my memory gaps by walking and speaking freely. Again, it was my luck that I trained the talk two times before with my notes visible and I knew most of the information already by heart. Only problem was that I had to read the slides too much, which caused me to face away from the audience. Whatever, I think the talk was much better, informative and funnier than most of the other talks I have seen. Seriously, some other people were filling their slides with info-junk to an unreadable level. Slides were good, Science was good, hacking was good. As usual, questions were answered with information and a Mozartkugel. Some pretty good questions by the audience. Intersting question about “entailment”, where I would rather use the term inference. Hm, which fits here more? We published Aperture this day and that is important. The virtual graph approach was nice in the first place, but didn’t scale so good anyhow. Chris Bizer suggested later, to collect implementations of virtual graphs, aperture adapters and other adapters somewhere. He insisted to do this collection language-independent, which is probably the way to go.
Patrick Stickler and his napkin idea
All in all, nice talk, and I had a light heart again, relieved by the weight of having something important to do now. During the rest of the day, people walked up to me and I had many interesting conversations about the topic. Later on, Patrick Stickler came (who wasn’t in the talk but liked the idea of aperture anyhow) and presented an architecture about distributed URIQA, looked very well-though. Also the presentation media, a paper napkin scribbled with architecture is the medium of choice when you want to start something like Compaq or bigger.
Single most nice event of the day was the TimBl keynote. Some people around me said that timbl sounds somewhat fuzzy in his talking style. May be, but he is one of the guys that had this idea on his mind since 1980 and he is a coder, so I trust his ideas because he has a faint knowledge how to code this on a global level. In his talk, he dissed the foaf community a little that they don’t use URIs to identify people. Hm, thats one of these things thats still open. TimBl now has a foaf file and luckily, he has a uri for himself, so we can identify him now using this uri. If this is good or bad, we will see. He also focused on people that don’t put rdf files at the place where their uris point. Yes, its 2005 and we have year ~five~ of Semantic Web and people are still too dumb to know what a URL is. Its a locator, and when you use http in your namespace, you have to put a document there. And when you use these URIs in your documents, expect people to download something from there. I nearly laughed my head of during this part of his talk, as it shows how blue-eyed some people are stepping towards the semantic web. Luckily, all my ontologies are at http://www.gnowsis.org/ont/ which I hope to host rather statically the next years. In the questions of the talk I asked if the uri crisis is solved, but I should have asked “what does a URI identify and what is the uri of Love?” but was too dumb to to it. So timbl answered, but I meant something else. Argh, leobard, you nincompoop. Whatever, we will find out how to identify abstract concepts sooner or later, its just a 3000 year old philosophical problem that will be solved by informatics in the 21st century. hm. Timbl also discouraged us to use arbitrary XYZ identifiers, namely lifescience identifiers, which sounds a good idea, we’ll have to see how we cope with it. During the talk, I suddenly realised that I miss all the good stuff by not being online, so I went online and found a lot of people sitting in the room being online and chatting about the talk.
So things to do for myself are: change my foaf file so that my URI points to the URI where I am. perhaps I should also move my foaf file to a saver place, where I can withstand the years. argh, uri crisis. Use a seperate domain, use a subfolder, use a file extension? hm. Timbl didn’t use a file extension in his uri, so this may be a good indicator.
In the evening, I was really tired. really. Having sat, sun, mon, tue, wed each day discussions and exchange of info wholeday until 3am in the morning, I was zombie mode. We went to dinner in the place where we were lunching on monday (the hotel next to the tourist office, nice food there). Just the DFKI members, nice talking. Later, we went to wrapup the week in the King’s Crown. Chris joined us and I started to have an uri crisis discussion with Sören. Bad idea, I was tired and emotional and pragmatic about this. During all years of uri crisis discussion, I never found a solution to it, so also not on this evening. Whatever, no one can give an answer on what a uri identifies, so we remain at the beginning there. I still wonder what is the right uri for concepts like “love”, “Leo Sauermann”, “Microsoft” or “ISWC2005”. Hope to find a solution for this age-old problem soon. Hope that god stops me discussing this one. We went sleeping early.