Greetings all, I'm pleased to report that we're back to work here at UMD, and the NLP group has resumed meetings after a year-long hiatus. We'll be meeting most Thursday afternoons at 2pm this semester. The "We" includes me, and first year graduate students Atul Kulkarni and Varada Kolhatkar.
Today we started off by discussing an article I recently wrote that is "in press" and tries to tie together a lot of work on recognizing similar contexts and measuring contextual similarity into one larger package....it's written for a non-technical audience and tries to use a lot of examples to show how many problems that sometimes appear quite different are really underneath it all very similar. This whole idea is something I've talked about at various times over the years, how things like WordNet-Similarity and SenseClusters are close cousins in some ways, for example, and I guess this article is trying to articulate that to some degree. You are welcome to download if you like, and of course comments are always appreciated. Computational Approaches to Measuring the Similarity of Short Contexts : A Review of Applications and Methods To appear in the South Asian Language Review (http://www.salr.net) http://www.d.umn.edu/~tpederse/Pubs/pedersen-salr-2007.pdf We'll be discussing this article again next Thursday Sept 20, where we will each come up with 2 or 3 NLP problems that can be solved by identifying similar contexts and that aren't already mentioned in the article. Ted -- Ted Pedersen http://www.d.umn.edu/~tpederse
