Within the field of artificial intelligence, researchers have been exploring the task of eliciting the preferences of users, towards the development of effective personal digital assistants for these users. This topic has been explored from different perspectives including: user modeling (the field of determining appropriate methods for representing a user and reasoning with that user model, as part of an agent's automated reasoning); constraint satisfaction (the field of determining a set of constraints that best represent a particular scenario and developing methods for determining the most effective solution for these constraints); decision-theoretic reasoning (the field of modeling the concepts of utility, benefit and cost, towards the proposal of effective methods for maximizing utility, for users); voting and auction settings, for multiagent systems (a field where reasoning about strategies for bidding, considering game theory, becomes important).

This course will begin with some lectures by the instructor, to cover background material, offering overviews of: artificial intelligence, user modeling, constraint satisfaction, decision-theoretic reasoning and voting. Then the majority of the class time will be spent having students do oral presentations of existing papers in the field of preference elicitation, in order to deepen our knowledge of the various approaches currently being proposed in this subfield.

We will then organize a group assignment, to brainstorm on the relationship between the different perspectives on preference elicitation (user modeling, constraint satisfaction, decision-theoretic and voting), with each group producing a document that presents new insights into the topic. We will set aside one entire class for the different groups to talk about their ideas and their proposals.

Students will also be asked to complete a major project on a topic relevant to the course. The project will require the analysis of existing work in the field as well as the development of some original ideas. Implementation-oriented projects are acceptable, providing that there is also a component where the student demonstrates capability in analyzing existing work in the field. Students will choose their own project topic and have it approved by the instructor.