David DeVault

Education

Research Interests

I am interested in building computer programs that can engage in natural, real-time, task-oriented conversations with human beings. This isn't possible in general yet; making it possible will require, among other things, that we develop better models of how language works in context, and find ways to realize substantive linguistic models in efficient computational implementations. I believe we will only succeed by an interactive approach to model building and implementation. In my research, I try to develop clear, principled, implementable models of fine-grained conversational phenomena and then investigate the viability of these models in implemented dialogue systems. My approach is to bring linguistic models into contact with computational techniques for generating and understanding utterances on the one hand, and into contact with real-world data from task-oriented human conversation on the other.

Together with my colleagues, I have built two dialogue systems: COREF [1,2,3,4,5] and FIGLET [1,2].

Publications

Making Grammar-Based Generation Easier to Deploy in Dialogue Systems David DeVault, David Traum, and Ron Artstein, The 9th SIGdial Workshop on Discourse and Dialogue (SIGdial 2008), Ohio, June, 2008.

Practical Grammar-Based NLG from Examples David DeVault, David Traum, and Ron Artstein, The Fifth International Natural Language Generation Conference (INLG 2008), Ohio, June, 2008.

Thoughts on FML: Behavior Generation in the Virtual Human Communication Architecture Jina Lee, David DeVault, Stacy Marsella, and David Traum, The First Functional Markup Language Workshop, Estoril, Portugal, May, 2008.

Managing ambiguities across utterances in dialogue David DeVault and Matthew Stone. The 2007 Workshop on the Semantics and Pragmatics of Dialogue (DECALOG 2007), University of Trento, Italy, May, 2007.

Enlightened update: A computational architecture for presupposition and other pragmatic phenomena Richmond H. Thomason, Matthew Stone and David DeVault. To appear in Donna Byron, Craige Roberts, and Scott Schwenter, eds., Presupposition Accommodation. For the Ohio State Pragmatics Initiative, 2006. *** This is a draft version (comments welcome!) ***

Scorekeeping in an Uncertain Language Game David DeVault and Matthew Stone. The 10th Workshop on the Semantics and Pragmatics of Dialogue (brandial 2006), University of Potsdam, Germany, September, 2006.

Societal Grounding is Essential to Meaningful Language Use David DeVault, Iris Oved, and Matthew Stone. Proceedings of the Twenty-First National Conference on Artificial Intelligence (AAAI-06), Boston, Massachusetts, July, 2006.

An Information-State Approach to Collaborative Reference David DeVault, Natalia Kariaeva, Anubha Kothari, Iris Oved, and Matthew Stone. ACL 2005 Proceedings Companion Volume. Interactive Poster and Demo Track., University of Michigan, June, 2005.

Interpreting Vague Utterances in Context David DeVault and Matthew Stone. The 20th International Conference on Computational Linguistics (COLING 2004), University of Geneva, Switzerland, August 23rd-27th, 2004.

Natural Language Generation and Discourse Context: Computing Distractor Sets from the Focus Stack David DeVault, Charles Rich, and Candace L. Sidner. 17th International FLAIRS Conference (FLAIRS-2004), Miami Beach, Florida, May, 2004. Also available as MERL technical report TR2004-004.

Domain Inference in Incremental Interpretation David DeVault and Matthew Stone. Fourth Workshop on Inference in Computational Semantics (ICoS-4), Nancy, France, September, 2003.

A flexible eyetracker for psychological applications. David DeVault and Alan Bond. Fifth IEEE Workshop on Applications of Computer Vision (WACV2000), Palm Springs, California, December, 2000.

Collaborations

I am currently a student in the VILLAGE at Rutgers, where I collaborate with Matthew Stone.

In addition to my work on dialogue systems, I am also working with Clinton Andrews to develop a multi-agent simulation of economic markets for niche products.

During the summer of 2003, I was an intern at Mitsubishi Electric Research Laboratories working on natural language generation in COLLAGEN collaborative agents with Chuck Rich and Candy Sidner.

As an undergraduate at Caltech, I worked with Alan Bond. Together we investigated eye movements as people evolved new strategies for solving the Tower of Hanoi puzzle, and designed a new eye tracker to make our experiments possible.

Contact Information

Email: ddevault AT cs DOT rutgers DOT edu

Mailing address:

Department of Computer Science
Rutgers, the State University of New Jersey
110 Frelinghuysen Road
Piscataway, NJ 08854