HuA Ai

Intelligent Systems Program 5420 Sennott Square University of Pittsburgh Pittsburgh, PA, 15260

Cell phone: (412) 780-1016 Email: hua AT cs.pitt.edu Website: www.cs.pitt.edu/~hua

Education

Ph.D. in Intelligent Systems, Summer 2009 (expected), University of Pittsburgh

M.S. in Intelligent Systems, 2006, University of Pittsburgh

B. S. in Computer Science, 2004, Shanghai Jiao Tong University, China

Research Interests

· User Modeling

· Affective Computing

· Spoken Dialog System Development

· Human Computer Interaction

· Natural Language Processing

· Machine Learning

Working Experience

Research Assistant University of Pittsburgh, PA

Intelligent Systems Program                                                                          December 2004 – current

I work with Dr. Diane Litman on the following research projects:

1.  User Simulation for Tutoring Dialog Systems

User simulation is a computer program which simulates human user behaviors using statistical user modeling techniques. Simulated users are used to automate and speed up the process of designing and developing spoken dialog systems. In my thesis research, I examine the three important factors of constructing user simulations. I also design controlled experiments to evaluate the simulations in different system development tasks.

2.  Using Lexical Features to Predict Student Emotions during Spoken Tutoring Dialogues

Recent trends in human computer interaction research emphasize the importance of detecting and adapting to different user emotions. In this study, we present an approach using lexical features to predict user emotions when users interact with a spoken dialog system. Salient words, which appear more often in one emotion category than in other emotion categories, are showed to be useful features in predicting user emotions using machine learning approaches.

Research Intern Bosch Research, CA

Human Machine Interaction Group                                                           September 2007 – April 2008

I worked with Dr. Fuliang Weng on In-Car Spoken Dialog System Development. This work was motivated by the needs of automatically evaluating a spoken dialog system quickly and repeatedly. We implemented a testing infrastructure in which evaluations with human users were replaced by experiments with user simulations. We also proposed a set of evaluation measures in the simulation context to synthesize objective and subjective scores of system performance using regression models.

Research Intern Institute of Creative Technologies at University Southern California, CA

Virtual Reality Research Group                                                                      May 2006 – August 2006

I worked with Dr. David Traum on using context features to improve dialogue move and parameter tagging in a spoken dialogue system. Context information is very useful in the task of understanding natural language. In this work, we used machine learning algorithms with context information to boost the performance of an interpretation module in a task-oriented spoken dialog system. We also used context related features to recover automatic speech recognition errors.

Research Assistant Shanghai Jiao Tong University, Shanghai

Natural Language Processing Group                                                         September 2002 – May 2004

I worked with Dr. Yongcheng Wang on a Chinese news question answering system. I worked on the natural language understanding module for this system which accepted queries in free forms. A user study was conducted to collect and rank the frequently used query patterns in order to identify keywords in user queries.

Publications

· Ai, H. and Litman, D. 2009. Setting Up User Action Probabilities in User Simulations for Dialog System Developmen. Proceedings 47th Annual Meeting of the Association for Computational Linguistics: Human Language Technologies (ACL). Suntec, Singapore.

· Ai, H. and Weng, Fuliang. 2008. User Simulation as Testing for Spoken Dialog Systems. In Proc. 9^th SIGDial Workshop on Discourse and Dialogue, Ohio, USA.

· Ai, H. and Litman, D. 2008. Assessing Dialog System User Simulation Evaluation Measures Using Human Judges. In Proc. 46^th ACL, Ohio, USA.

· Ai, H.; Raux, A.; Bohus, D.; Exkenazi, M.; and Litman, D. 2007. Comparing Spoken Dialog Corpora Collected with Recruited Subjects versus Real Users. In Proc. 8^th SIGDial Workshop on Discourse and Dialogue, Antwerp, Belgium.

· Ai, H. and Litman, D. 2007. Knowledge Consistent User Simulations for Dialog Systems. In Proc. Interspeech 2007, Antwerp, Belgium.

· Ai, H.; Roque, A.; Leuski, A.; and Traum, D. 2007. Using Information State to Improve Dialogue Move Identification in a Spoken Dialogue System. In Proc. Interspeech 2007, Antwerp, Belgium. [poster]

· Hua Ai, Joel R. Tetreault, and Diane J. Litman. 2007. Comparing User Simulation Models for Dialog Strategy Learning. In Proc. NAACL-HLT, Rochester, NY. [poster]

· Roque, A.; Ai, H.; and Traum, D. 2006. Evaluation of an Information State-Based Dialogue Manager. Brandial 2006: The 10th Workshop on the Semantics and Pragmatics of Dialogue. Potsdam, Germany.

· Ai, H.; Litman, D.; Forbes-Riley, K.; Rotaru, M.; Tetreault, J.; and Purandare, A. 2006. Using System and User Performance Features to Improve Emotion Detection in Spoken Tutoring Dialogs. Proceedings of Interspeech 2006 ICSLP. Pittsburgh, PA (to appear in September).

· Ai, H. and Litman, D. 2006. Comparing Real-Real, Simulated-Simulated, and Simulated-Real Spoken Dialogue Corpora. Proceedings of the AAAI Workshop on Statistical and Empirical Approaches for Spoken Dialogue Systems. Boston, MA.

· Ai, H.; Harris, T.; and Rose, C. P. 2006. The Effect of Miscommunication Rate on User Response Preferences. CHI, Work-In-Progress.  Montreal, Canada. [poster]

Presentations

· “User Simulation for Spoken Dialog System Development”. Speech Lunch, Stanford University, Stanford, CA, February 2008.

· “Comparing User Simulation Models for Dialog Strategy Learning”. AI Forum, University of Pittsburgh, Pittsburgh, PA, February 2007.

· “Comparing Real and Simulated Spoken Dialogue Corpora”, “Dialogs on Dialogs” Group Meeting, Carnegie Mellon University, Pittsburgh, PA, March 2006.

· “Using Lexical Features to Predict Student Emotions during Spoken Tutoring Dialogues”. AI Forum, University of Pittsburgh, Pittsburgh, PA, March 2005.

Awards

· 2008-2009, Mellon Fellowship, University of Pittsburgh.

· 2005-2006, FSA Fellowship, Intelligent Systems Program, University of Pittsburgh.

· 2004-2005, FSA Fellowship, Intelligent Systems Program, University of Pittsburgh.

· 2003, Renmin Scholarship, Shanghai Jiao Tong University.

· 2001, Renmin Scholarship, Shanghai Jiao Tong University.

Activities

· Organizer of the 4th Young Researchers Roundtable on Spoken Dialogues Systems Workshop, 2008.

· Reviewer for IEEE Transactions on Audio, Speech and Language Processing.

· Reviewer for the 22nd International Florida Artificial Intelligence Research Society Conference.

Computer Skills

· Programming Languages: C++, Java, Perl, SQL.

· Application Software: Weka, SPSS, Matlab.

· Operating Systems: Windows XP/2000/NT, UNIX/LINUX.

References

Available Upon Request.