jump to navigation

Work sets you free October 11, 2006

Posted by Sharath Rao in CMU, ideas, life.

I have often been asked by friends about what is it I am working on or what exactly I mean when I say “doing research”. Some of my common replies have been a) Speech Translation b) Mathematics and Statistics c) “Don’t ask man, lets talk anything except work”. This is ironic because graduate students and faculty members love to talk about their work and are given to think that what they do is really so important that every person on this earth must know about it. I have no idea why I haven’t been so excited about talking about things I work on; definitely not because its anything defense sensitive ( if it were, an Indian National wouldnt have been working on it ! ). So this post hopefully makes amends.

I am at a researchers’ gathering here in White Plains, about 25 miles North of downtown New York City. This is a meeting of about 250 researchers in the field of Speech Recognition, Artificial intelligence and Natural Language Processing. This is a part of what is called the GALE project, a project funded by the Defense advanced research projects agency (DARPA ). ( One of DARPA’s project lead to the birth of the internet ). This is also the project that funds my education, i.e. my tuition and stipend comes from the money that is allocated to this project. Now here is what the GALE project is about, right off their website.

The goal of the GALE (Global Autonomous Language Exploitation) program is to develop and apply computer software technologies to absorb, analyze and interpret huge volumes of speech and text in multiple languages, eliminating the need for linguists and analysts and automatically providing relevant, distilled actionable information to military command and personnel in a timely fashion. Automatic processing “engines” will convert and distill the data, delivering pertinent, consolidated information in easy-to-understand forms to military personnel and monolingual English-speaking analysts in response to direct or implicit requests.

To strip off the jargon, here is what you would do. You are watching an Arabic speech given by Saddam Hussein on TV. You know only English, don’t understand Arabic but have a question about the speech he just made. Now you are required to design a system such that you should be able to type your question in English and the system should be able to answer your question (obviously in English) !! All of this should be completely automatic – at no stage should there be any human involved!! This means the system should first convert Arabic Speech to Arabic text (Speech Recognition), translate this Arabic text to English text (Speech Translation) and then use Google like information retrieval techniques to extract the answer to your question. (if it exists)

Ofcourse this is a huge project with over 300 researchers all over 15 research labs in North America and Europe working on it. The research community has been grouped into 3 teams, each of which have 4-5 research labs. Carnegie Mellon University is grouped with IBM, Stanford University, University of Maryland College Park, Johns Hopkins and Brown University.

I am working on two parts of this huge system – on improving the translation accuracy by breaking sentences into meaningful and manageable parts and then cleaning up the this text by removing disfluent utterances. This involves whole of statistical analysis and no knowledge of language itself, after all its automatic. I know neither Arabic nor Chinese and for me these are mere mathematical symbols that need to be manipulated.

So I was here to give a talk ( mine was one among 25 other talks ) describing the work we have been doing in the past few months at Carnegie Mellon. It has been quite an occasion, being perhaps the youngest in that hall of 250 people with a good number of people working in this field even before I was born. So in some sense it was still among the most important talks I ever gave and it went off quite okay. No, dont let this generate responses like “Man, you are going places” (like a friend remarked), there was not anything outstanding about the work itself and I would have been far happier to have got better results. But 2 years back when I was applying to universities in the United States, if someone had offered me this view of the future, I would have gladly taken it.

There wasn’t a better place to turn 25.

P.S: It appears that the last few days have seen some kind of a blogging high. I still have enough material to write atleast 5 different posts, just no energy. Maybe I should get myself one of these.

P.P.S: The title of this post, if readers can connect the dots, is rather gross.



1. Sadness in San Francisco « Epistles - March 21, 2007

[…] I got here 24 hours ago in connection with the DARPA project meeting ( like the kind I attended in Oct 2006 at NYC ) and will be here for another 5 days. And unless something life-changing happens in these 5 days […]

Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out / Change )

Twitter picture

You are commenting using your Twitter account. Log Out / Change )

Facebook photo

You are commenting using your Facebook account. Log Out / Change )

Google+ photo

You are commenting using your Google+ account. Log Out / Change )

Connecting to %s

%d bloggers like this: