home · mobile · calendar · colloquia · 2007-2008 · 

Colloquium - Ponte

Machine Translation
Jay Ponte
Google, Inc.

Thanks to the internet, the ability to communicate over long distances, across social groups, in spite of political boundaries and without regard to time zones has never been better but one barrier that persists is language.

Machine Translation services on the Internet are used by millions of people every day and while the quality often leaves much to be desired, many users find these translations useful and even essential. If we can significantly improve the quality of MT, it will enable communication and information exchange across language boundaries on a scale never imagined even a few years ago.

At Google, we started a research project to do exactly that in 2003. By 2005, we were participating in the annual MT evaluation conducted by NIST and, in 2006, we launched our world class translation system on Google.com. In this talk, we will look at the challenges of building a world class MT system and making it available to the world and at recent work to improve translation quality.

Jay Ponte received a BS with honors in computer science from Northeastern University in 1993. He received MS and PhD degrees in computer science from the University of Massachusetts at Amherst in 1996 and 1998, respectively. While in graduate school, he worked on classification of text-based medical records, Chinese natural language processing, topic segmentation, and information retrieval under the direction of Professor W. Bruce Croft. His dissertation work in probabilistic language modeling for information retrieval has been influential in the information retrieval field. After graduate school, he joined GTE Laboratories to start up the SuperPages Advanced Development Group and worked in the areas of web search and text classification. In 2000, he joined the MITRE Corporation where he continued to work on probabilistic approaches to natural language processing and information retrieval and also on software architectures in support of these technologies. As of 2002, when not writing about himself in the third person, he works for Google Inc. in the area of statistical language translation, specifically on word alignment models, phrase based translation models and end-to-end system training infrastructure.

This talk is sponsored by the Computer Science Department ACM Student Chapter.

Department of Computer Science
University of Colorado Boulder
Boulder, CO 80309-0430 USA
May 5, 2012 (14:13)