Research Topics

Case-Based Preference Elicitation: UAI'98, UAI'01, AIJ'03
Preference Elicitation via Theory Refinement: IUI'01, AAAI'02, JMLR'03
Constraint-Based Preference Elicitation: UAI'97, UAI'99
Abstraction-Based Probabilisitic Planning: UAI'96, KR'98, AMAI'98
Generalized Semi-Markov Decision Processes: HICSS'03
Theorem proving (PVS): ICSE'04
Interval Probabilities, Interval Bayesian Networks: AAAI'97, KR'98, AMAI'98
User Modeling: IUI'01

PhD Dissertation: Reasoning with Partial Preference Models (PDF 518KB, 65 pages)

Classical decision theory provides a normative framework for representing and reasoning with complex preferences. Straightforward application of this theory to automate decision making is difficult due to the high cost of eliciting preferences. The objective of this thesis is to develop a flexible decision-theoretic framework for eliciting and reasoning with preferences and to apply the tools and techniques of this framework to build a practical decision support system. This framework incorporates two orthogonal, complementary approaches to eliciting and reasoning with partial preference information. The first approach is grounded in classical multi-attribute utility theory, and can make effective use of qualitative preferential statements represented by logical comparative sentences. The second approach applies ideas from case-based reasoning and collaborative filtering to address the elicitation problem. The theoretical results of this framework have been tested extensively in a number of experiments involving real-world data, as well as found an application in a recommender system.

Selected Papers