SmILE Semantic Information Systems and Language Engineering Group

With a short break after trips to IJCAI'09 and AIME'09, I finished the summer conference marathon at CBMS'09, 22nd in the series of IEEE symposiums on Computer-Based Medical Systems. The conference was rather small, organised through two days as one general session and a lot of parallel special tracks with more or less independent program committees. Such a format was rather unfamiliar for me, but apparently it is a sort of cross between usual computer science conferences and biomedical symposiums. Once again in this summer's conference chain, the community was a bit new to me (as I was to the community itself), although at the first lunch I met at least one guy I knew before from the Elsevier Grand Challenge - George Thoma, a group leader at National Library of Medicine.

The topics ranged from medial imaging and image recognition, through data mining, medical knowledge representation and modeling, to grid computing in the life sciences domain. A lot of emphasis, both within the two invited talks and across the regular conference sessions, was put on the image analysis. Also, the application domains much more often regarded clinical data (while AIME'09, the other "computer/life science" conference I attended recently, was more or less equally about processing of both scientific/experimental and medical/clinical data). For a guy coming from the Semantic Web community, it was quite pleasing to see a couple of papers attempting something like an early adoption of the semantic technologies (i.e., RDF, ontologies or reasoning) in the realm of real-world medical data, without any Manchester people directly or indirectly pushing them to use OWl and stuff ;) . If you're interested in anything particular from the list of accepted papers, just come to me and I'll try to give you an overview plus the paper (if I'll manage to get the USB stick with the proceedings working again after I carelessly put it in a weird Windows system which caused it not to mount on my laptop anymore). I only have to warn you beforehand - most papers from the proteomics session I attended (hoping I'll learn something useful about analysis of proteomics data) are not worth reading at all. It was one of the funniest sessions I've ever attended - one guy, incidentally co-chair of the session, presented 3 out of 6 papers, the second co-chair presented one more and then there was only one paper left to be presented by someone else, since another author didn't make it on time... In the beginning I was even afraid that it will be solely a one man show with presentations of rather mediocre papers (in the end, only two out of the five presentations made sense scientifically and addressed an interesting problem as far as I could tell). But the other sessions I went to were fortunately not as bad as this one...

My talk was in the Knowledge Discovery and Decision Systems in Biomedicine track and it was basically about CORAAL, similarly to my AIME'09 talk, although this time there were few more technical details covered, too (see the slides if interested). Once again, I fortunately managed to give a reasonably good talk, as reflected by a couple of people from the audience later on. People asked some quite unsurprising questions related to the stuff that deliberately wasn't explained in too much detail in the slides. There was also one interesting remark from an Australian oncologist regarding problems with automated extraction or inference of non-explicit common-sense knowledge from expert texts, which seems to be quite important for their use cases. We discussed that a bit also after the talk and hopefully there will be some cooperation later on (as he was interested in trying our stuff on their data when I'll make it available in a usable enough shape). I've also talked a bit with a girl from University College, London, who is doing quite an interesting research in concept-based queries for medical data within a rather large UK project developing the ONIX oncology information exchange platform. Possibly we will combine our research later on and try to do something interesting, maybe even as a plug-in for the ONIX service if it will be successful and mature enough at some stage.

Wrapping up, the conference was rather interesting from my personal subjective point of view, although perhaps not as packed with exciting and relevant research as AIME or IJCAI I visited before. However, it was definitely much cooler than any other conference I've ever attended - literally. Given the fact it was located in Albuquerque, i.e., in the middle of the high desert in New Mexico, I don't really find very reasonable setting up the air conditioning in the conference rooms 20 degrees below the outside temperature. After enjoying the desert air flowing into my room through the open balcony all the night, I felt like in a freezer when I came to the conference downstairs. Should this experience be deemed statistically relevant, I can only recommend to bring a sweater and thick socks whenever you're going to a conference in a US desert area ;) .