(Late and Boring) Report on ACM Multimedia 2001, Ottawa, Canada September 30-October 5, 2001 Report by Lynda, 28 December 2001 To be found as http://www.cwi.nl/~media/trip-reports/acm-mm2001.txt Lloyd, Frank and I went to ACM MM01. I was not inspired - although perhaps Lloyd and Frank were?? There was only one keynote (Ramesh Jain) and one panel (Lloyd is on the panel) at the conference. There was a demo room. There seems to be no Web and no semantics. And there wasn't any hypertext to start with (although people talk about interaction). Tuesday Ramesh Jain's keynote mentioned all the problems the Semantic Web promises to solve. Ralf Steinmetz asked the question as to whether a knowledge-based approach would work. Ramesh answered using the ontology word - but he is not an AI person. He probably can't answer the question. The AI world doesn't understand MM. Confirms we are on the right track but we're on our own. Some interesting people were there: Wolfgang Klaus, Susanne Boll, Jo Konstanz, Brian Bailey, Anna Benitez. (Although few were to be found at the paper sessions...) (Larry Rowe went to Lloyd's SMIL tutorial - for a second time!) MPEG7 is around in the corridors, but not in the conference. Are there relevant papers somewhere else? An IBM paper (p117) was on summarizing videos (based on user interaction). What I didn't get is whether the "semantics" of the summary is emergent or whether domain knowledge is used. (I think the former.) The p127 paper did include domain knowledge - what was it & what was its representation? I'm not sure of the benefit of the coloured time/space representations, but they are cute. In the evening, chatted to Lloyd, Rob and Pablo(?) about the SYMM working group. It was interesting. I'm not sure what I learned that was new. The issue that was troubling me was the openness of working groups to non W3C members. This seems to have something to do with RAND (Reasonable And Non-Discriminatory) and RF (Royalty Free). The main fear is that someone external to W3C could take out a patent on ideas discussed in the open working-group mailing list. Voting stays restricted to members. Wednesday First session was "Lloyd's" panel "Is streaming media mainstream". Lloyd gave the most inspired talk. Message was that streaming media will only become mainstream when "normal people" have learned to make video and audio content of themselves - like we can just about take photographs. Brian Bailey gave a very nice paper presentation. I thought it deserved a best student paper award, if not mention. Anyway, his presentation complimented the paper very nicely, and is available on DVD (which should have arrived by now I suspect). Saw Anna's demo. Combining video and wordnet (more or less) for retrieval. Should send her Guus's ref, and copy her stuff to Guus. I was inspired by the "singing" demo. By singing the melody line of a tune (and it didn't have to be the first line) the system searched 10,000 songs and found the one you were singing. I'm not quite sure what sort of analysis they did, but it must have been pitch and rythmn independent. I sung the middle bit of the Penny Lane chorus, and the system found the right tune. Met Ed Hartley and Adam Lindsey, of Lancaster and MPEG7, and chatted at dinner. Thursday The talk from Virage was interesting because it was a case of MM research going to product. Although I think it was 1993/4 research. Basically video plus features in a database. I listened to a doctoral consortium presentation from Mrelle(?)steinm(?) from Twente. It was very good. We should have her over sometime to give a talk. Very HCI and less MM. (I tried looking for her on the Web, and couldn't find her.) I chatted with Wolfgang Klas about BRIDGE, Susanne, GMD-IPSI and European funding in general. It was very interesting actually. We were also chatting with some annoying guy, but he made me state things explicitly about our research, e.g. MPEG7 seems too complex. XML is "too complex" when compared with the first HTML, but after everyone understood HTML they were prepared to accept the more complex step of XML. Same will be true of the semantic stuff, It will take longer to acceptance though. (Dick is still not convinced about semantics - not sure if we will be able to convince him, but then, he never liked hyperlinks either... :-) -------