Description: image001.jpg

Anatole Gershman

Distinguished Career Professor of Computer Science

Language Technologies Institute

6415 Gates-Hillman Complex

Carnegie Mellon University

Pittsburgh, PA 15213-3890

 

Email:

Description: image002.jpg

Phone: (412) 268-8259

Fax:        (412) 268-6298

 

 

Research Interests

       Information extraction and co-reference resolution

       News personalization

       Knowledge and language acquisition through active learning

       Reasoning under uncertainty

       Fusion of information from disparate sources

       Making everyday objects smart

Teaching

2007-2012:  Inventing the Future of Services

Fall 2014: 15-291 Applied Computational Intelligence Lab

Research Projects

Recent:

Analysis of Uncertain Data (IEEE International Conference on Systems, Man, and Cybernetics, 2009 PDF, PDF)

Encore (Entity Co-reference Resolution)

News Personalization (PDF)

Metal – Multi-lingual Metaphor Identification and Interpretation (ACL2014 PDF)

Current:

Contextualized News Summarization (IEEE IS14 PDF, SIGIR PIR 2014 PDF, SIGIR 2013 PDF, LREC12 PDF)

TREE – Teachable Relations and Event Extractor

Analysis of Customer Service phone calls

Short Bio

My entire professional career has been dedicated to a simple question: How can we use Information Technology to act more intelligently? In a commercial setting, where I spent most of my professional life, I had to apply this question to activities that make business sense such as creation of new productivity tools, products and services. Every advance in basic technologies created opportunities for intelligent business applications.  In the 1980s, I led the design and implementation of the first production-quality automated Natural Language Processing system that handled international money transfer telexes for Citibank. When I joined Accenture in 1989, I created a lab that explored the use of multimedia for knowledge management and training. We also built early, pre-web prototypes of interactive multimedia-based eCommerce applications. In 1995, we released the worlds first eCommerce agent that shopped the web for music CDs.

While the whole world was falling in love with eCommerce and predicting the demise of brick and mortar, we quickly realized that the physical world was not about to disappear.  We coined the term uCommerce for Ubiquitous Commerce and built a number of demonstrations of services that use the combination of personal mobile devices and the surrounding environments. As we were building early examples of Ubiquitous Commerce, it became clear to us that these services would require complex automated intelligence based on detailed models of human behavior. Automated creation of such models requires an ability to fuse often ambiguous and noisy information from disparate sources, including physical sensors and the web. As a result, we created an Analytics and Intelligence group at Accenture Technology Labs which now includes several CMU graduates.

My professional career started in Natural Language processing at Yale University where I got my Computer Science Ph.D. in 1979. More recently, I realized that the vision of ubiquitous intelligent services will not be fulfilled without significant advances in Language Technologies. At the end of 2006 I retired from my position as Global Director of Research and Chief Scientist at Accenture and joined the LTI faculty in April 2007.