Very exciting news Dr. Patwardhan, and very hearty congratulations on defending your PhD dissertation, and also the new position at IBM. This is a very intriguing project, and one that will certainly be followed with interest by the general public (in large part due to the popularity of Jeopardy). It's also a very interesting melding of supercomputing with an NLP problem - one interesting question will be where the balance will fall between simply using really powerful (and simple) data driven techniques versus more sophisticated NLP methods.
In the case of Deep Blue (IBM's chess playing machine about a decade ago) my impression is that there was a large amount of chess knowledge built into the system, and then special purpose hardware was made to search the very large problem space that chess represents. There are some similarities between chess and NLP I guess, at least with respect to impossibly large search spaces, although of course NLP has many more rules (most of which are violated regularly). So one wonders where the balance will fall - lots of high powered data driven methods, or lots of complicated rules. I suspect the answer will be a bit of both, along with some magic dust to make it all come together. :) Thanks again for the update, and best of luck as you move up north yet again. Do stay in touch! Ted On Thu, Sep 24, 2009 at 6:07 PM, Siddharth Patwardhan <[email protected]> wrote: > > > > Hi all, > > I haven't been active on this list for quite a while. I'd like to first > introduce myself to Ted's new students. Several years back, I was Ted's > student at UMD, working on my MS thesis research in NLP -- specifically, > word sense disambiguation and semantic relatedness of words and > concepts. Since graduating from UMD, I have been pursuing at PhD at the > University of Utah under the guidance of Dr. Ellen Riloff. I have > continued to lurk on this list over the years, and have followed > various discussions and achievements of students in this group. > Occasionally, I send updates to Ted, and I thought I'd share this latest > piece of news with the group as well: > > I am happy to report that I successfully defended my PhD dissertation > last week. I've been working on event-based information from text, and > for my dissertation research I devised an approach based on "event > recognition" for improving information extraction accuracy. I am still > in the process of finalizing my dissertation, and I will be sure to send > a link to the dissertation on this list once I have it done. > > I would like to express my gratitude to many of you in the NLP group at > UMD for your support through this process. I would like especially thank > Ted, Bano, Saif, Bridget and Amruta... you all have been extremely > supportive, and I am very grateful for your support! > > In other related news, I just accepted an offer of employment at the IBM > T. J. Watson Research Center as a post-doctoral research associate. I > will be working on the Jeopardy challenge: > > http://www.research.ibm.com/deepqa/index.shtml > > http://www.nytimes.com/2009/04/27/technology/27jeopardy.html?_r=1&scp=1&sq=ibm%20jeopardy%20challenge&st=cse > > I will be starting work at IBM in the first week of October, and I'm > looking forward to it. I remember there was some discussion about the > Jeopardy challenge on this list a few months back. I will keep you all > updated on the progress on this and on the DeepQA project. > > Hope to see you all soon. > > -- Sid. > > -- Ted Pedersen http://www.d.umn.edu/~tpederse
