SIGIR 2007 August 3, 2007Posted by Andre Vellino in Data Mining, Information retrieval, Statistical Semantics.
I’ve just returned from SIGIR 2007. As you’d expect, the participants and presenters at this conference this year were dominated by search-engine research labs: Google, Yahoo, Microsoft, Ask, Baidu.
There’s no doubt that the lure of financial reward is drawing the best minds in Information Retrieval away from academic institutions and into the big search-engine companies. One has to worry about this trend. Even though there is a lot of collegiality between individual researchers from different corporations, there are also market pressures that make novel IR techniques proprietary. This can’t be good for intellectual freedom or the advancement of knowledge.
The conference was dedicated to the memory of Karen Sparck Jones who died in April 2007, much to the chagrin of the IR community. She gave a recorded video talk of her Athena Award Lecture in which she muses speculatively about the broader future of IR in computing. She envisions a future in which computing devices will have an “IR Layer”, much as there is now an “IP Layer” and an “OS Layer” – i.e. standard APIs to perform basic IR tasks (classification / indexing / searching / stemming etc.) out of which IR applications could be built in a standard way. I think she may be right that this is what typically happens when a research field rich with application potential has matured.
The best student paper was, surprisingly, on the application of support vector machines for anti-spam (“Relaxed Online SVMs for Spam Filtering” by D. Sculley, G. M. Wachman (Tufts University)). It’s surprising because spam is such an un-sexy topic these days, but it deservedly won the award: very well laid out, good results, very well presented.
The best paper award was given to “Neighborhood Restrictions in Geographic IR“. Although it was a good paper, I just don’t see that it addresses a real, meaningful or important problem. My vote goes to a more theory-focused paper: “Estimation and Use of Uncertainty in Pseudo-relevance Feedback“. (It isn’t just because the paper was written at CMU that I liked it :-).)
Also from CMU was a Question-Answering paper entitled “Structured retrieval for Question Answering“. It was one of the few papers that used NLP, which I predict will make a comeback – especially for question-answering and knowledge extraction.
There were some entertaining social events. The one-man video-music show – MondoLeone – was quite entertaining, especially the story of his electric bass guitar which used to teach the “most unfamous famous bass player”. Another event was the banquet (with a couple of professional ballroom dancers playing to live Argentinean music – tango etc.) There are some 200+ pictures of the whole conference.