About Me

My work at Microsoft has been focused on topics at the intersection of information retrieval, natural language processing and machine learning. I work in the Text Mining, Search and Navigation (TMSN) group in Microsoft Research. My current interests include aspects of targetted/topical search, human computation games and search relevance. I have recently done some work on improving advertising relevance.

I am the Editor of SIG-IRList, a moderated newsletter from the ACM Special Interest Group on Information Retrieval.

I joined Microsoft and MSN Search as a Researcher in 1998, and came up with several innovative search features that were deployed in MSN Search. After some years in the Natural and Interactive Services Division and the Natural Language Group working on text mining and other areas, I worked in a small incubation group in Microsoft Research on news dissemination. In 2007, I spent several months in Microsoft Research Asia, Beijing, working on search relevance and search experience satisfaction. Since then, I have been in the TMSN group at Microsoft Research Redmond.

I got my PhD in Computer Science from the Tata Institute of Fundamental Research (TIFR). I worked at TIFR and at the National Center for Software Technology (NCST), Bombay till 1995, with a brief visiting stint at the Center for Machine Translation at Carnegie Mellon University, Pittsburgh. I helped create the Knowledge Based Computer Systems group at NSCT and was a founding editor of Vivek, A Quarterly in Artificial Intelligence. From 1995 to 1998, I worked at the Institute for Research in Cognitive Science at the University of Pennsylvania, Philadelphia, mainly on using computational linguistics to improve information retrieval.

Page Hunt

Try a game that we've been working on: Page Hunt, available at http://pagehunt.msrlivelabs.com

This game is like search in reverse: you're shown a web page, which you have to ‘hunt down' using queries sent to Live Search.

When you see the web page, think of a set of terms that can get back this page. Type these words into a search box. Page Hunt shows the top 5 results for this query from Bing. You get points if the web page you're 'hunting' is one of these top 5 result pages.

If you don’t get it right, change the query and try hunting again. If you get it right, you advance to the next page.

The game uses Microsoft Silverlight, so you will be prompted to load Silverlight when you first run it.

We hope Page Hunt is fun, something you’ll keep coming back to!