Abstract:
While decision theory provides an appealing normative framework for representing rich preference structures, eliciting utility or value functions typically incurs a large cost. For many applications involving interactive systems this overhead precludes the use of formal decision-theoretic models of preference. Instead of performing elicitation in a vacuum, it would be useful if we could augment directly elicited preferences with some appropriate default information. In this paper we propose a case-based approach to alleviating the preference elicitation bottleneck. Assuming the existence of a population of users from whom we have elicited complete or incomplete preference structures, we propose eliciting the preferences of a new user interactively and incrementally, using the closest existing preference structures as potential defaults. Since a notion of closeness demands a measure of distance among preference structures, this paper takes the first step of studying various distance measures over fully and partially specified preference structures. We explore the use of Euclidean distance, Spearmans footrule, and define a new measure, the probabilistic distance. We provide computational techniques for all three measures.

Abstract:
We investigate the application of classification techniques to utility elicitation. In a decision problem, two sets of parameters must generally be elicited: the probabilities and the utilities. While the prior and conditional probabilities in the model do not change from user to user, the utility models do. Thus it is necessary to elicit a utility model separately for each new user. Elicitation is long and tedious, particularly if the outcome space is large and not decomposable. There are two common approaches to utility function elicitation. The first is to base the determination of the users utility function solely ON elicitation OF qualitative preferences.The second makes assumptions about the form AND decomposability OF the utility function.Here we take a different approach: we attempt TO identify the new USERs utility function based on classification relative to a database of previously collected utility functions. We do this by identifying clusters of utility functions that minimize an appropriate distance measure. Having identified the clusters, we develop a classification scheme that requires many fewer and simpler assessments than full utility elicitation and is more robust than utility elicitation based solely on preferences. We have tested our algorithm on a small database of utility functions in a prenatal diagnosis domain and the results are quite promising.

Abstract:
Structured utility models are essential for the effective representation and elicitation of complex multiattribute utility functions. Generalized additive independence (GAI) models provide an attractive structural model of user preferences, offering a balanced tradeoff between simplicity and applicability. While representation and inference with such models is reasonably well understood, elicitation of the parameters of such models has been studied less from a practical perspective. We propose a procedure to elicit GAI model parameters using only "local" utility queries rather than "global" queries over full outcomes. Our local queries take full advantage of GAI structure and provide a sound framework for extending the elicitation procedure to settings where the uncertainty over utility parameters is represented probabilistically. We describe experiments using a myopic value-of-information approach to elicitation in a large GAI model.

Abstract:
User preferences for automated assistance often vary widely, depending on the situation, and quality or presentation of help. Developing effectivemodels to learn individual preferences online requires domain models that associate observations of user behavior with their utility functions, which in turn can be constructed using utility elicitation techniques. However, most elicitation methods ask for users' predicted utilities based on hypothetical scenarios rather than more realistic experienced utilities. This is especially true in interface customization, where users are asked to assess novel interface designs. We propose experiential utility elicitation methods for customization and compare these to predictivemethods. As experienced utilities have been argued to better reflect true preferences in behavioral decision making, the purpose here is to investigate accurate and efficient procedures that are suitable for software domains. Unlike conventional elicitation, our results indicate that an experiential approach helps people understand stochastic outcomes, as well as better appreciate the sequential utility of intelligent assistance.

Abstract:
This paper introduces CLEO, a novel preference elicitation algorithm capable of recommending complex objects in hybrid domains, characterized by both discrete and continuous attributes and constraints defined over them. The algorithm assumes minimal initial information, i.e., a set of catalog attributes, and defines decisional features as logic formulae combining Boolean and algebraic constraints over the attributes. The (unknown) utility of the decision maker (DM) is modelled as a weighted combination of features. CLEO iteratively alternates a preference elicitation step, where pairs of candidate solutions are selected based on the current utility model, and a refinement step where the utility is refined by incorporating the feedback received. The elicitation step leverages a Max-SMT solver to return optimal hybrid solutions according to the current utility model. The refinement step is implemented as learning to rank, and a sparsifying norm is used to favour the selection of few informative features in the combinatorial space of candidate decisional features. CLEO is the first preference elicitation algorithm capable of dealing with hybrid domains, thanks to the use of Max-SMT technology, while retaining uncertainty in the DM utility and noisy feedback. Experimental results on complex recommendation tasks show the ability of CLEO to quickly focus towards optimal solutions, as well as its capacity to recover from suboptimal initial choices. While no competitors exist in the hybrid setting, CLEO outperforms a state-of-the-art Bayesian preference elicitation algorithm when applied to a purely discrete task.

Abstract:
Complexity theory is a useful tool to study computational issues surrounding the elicitation of preferences, as well as the strategic manipulation of elections aggregating together preferences of multiple agents. We study here the complexity of determining when we can terminate eliciting preferences, and prove that the complexity depends on the elicitation strategy. We show, for instance, that it may be better from a computational perspective to elicit all preferences from one agent at a time than to elicit individual preferences from multiple agents. We also study the connection between the strategic manipulation of an election and preference elicitation. We show that what we can manipulate affects the computational complexity of manipulation. In particular, we prove that there are voting rules which are easy to manipulate if we can change all of an agent's vote, but computationally intractable if we can change only some of their preferences. This suggests that, as with preference elicitation, a fine-grained view of manipulation may be informative. Finally, we study the connection between predicting the winner of an election and preference elicitation. Based on this connection, we identify a voting rule where it is computationally difficult to decide the probability of a candidate winning given a probability distribution over the votes.

Abstract:
Decision theory has become widely accepted in the AI community as a useful framework for planning and decision making. Applying the framework typically requires elicitation of some form of probability and utility information. While much work in AI has focused on providing representations and tools for elicitation of probabilities, relatively little work has addressed the elicitation of utility models. This imbalance is not particularly justified considering that probability models are relatively stable across problem instances, while utility models may be different for each instance. Spending large amounts of time on elicitation can be undesirable for interactive systems used in low-stakes decision making and in time-critical decision making. In this paper we investigate the issues of reasoning with incomplete utility models. We identify patterns of problem instances where plans can be proved to be suboptimal if the (unknown) utility function satisfies certain conditions. We present an approach to planning and decision making that performs the utility elicitation incrementally and in a way that is informed by the domain model.

Abstract:
Preference queries incorporate the notion of binary preference relation into relational database querying. Instead of returning all the answers, such queries return only the best answers, according to a given preference relation. Preference queries are a fast growing area of database research. Skyline queries constitute one of the most thoroughly studied classes of preference queries. A well known limitation of skyline queries is that skyline preference relations assign the same importance to all attributes. In this work, we study p-skyline queries that generalize skyline queries by allowing varying attribute importance in preference relations. We perform an in-depth study of the properties of p-skyline preference relations. In particular,we study the problems of containment and minimal extension. We apply the obtained results to the central problem of the paper: eliciting relative importance of attributes. Relative importance is implicit in the constructed p-skyline preference relation. The elicitation is based on user-selected sets of superior (positive) and inferior (negative) examples. We show that the computational complexity of elicitation depends on whether inferior examples are involved. If they are not, elicitation can be achieved in polynomial time. Otherwise, it is NP-complete. Our experiments show that the proposed elicitation algorithm has high accuracy and good scalability

Abstract:
Combinatorial auctions where agents can bid on bundles of items are desirable because they allow the agents to express complementarity and substitutability between the items. However, expressing one's preferences can require bidding on all bundles. Selective incremental preference elicitation by the auctioneer was recently proposed to address this problem (Conen & Sandholm 2001), but the idea was not evaluated. In this paper we show, experimentally and theoretically, that automated elicitation provides a drastic benefit. In all of the elicitation schemes under study, as the number of items for sale increases, the amount of information elicited is a vanishing fraction of the information collected in traditional ``direct revelation mechanisms'' where bidders reveal all their valuation information. Most of the elicitation schemes also maintain the benefit as the number of agents increases. We develop more effective elicitation policies for existing query types. We also present a new query type that takes the incremental nature of elicitation to a new level by allowing agents to give approximate answers that are refined only on an as-needed basis. In the process, we present methods for evaluating different types of elicitation policies.

Abstract:
We state the problem of inverse reinforcement learning in terms of preference elicitation, resulting in a principled (Bayesian) statistical formulation. This generalises previous work on Bayesian inverse reinforcement learning and allows us to obtain a posterior distribution on the agent's preferences, policy and optionally, the obtained reward sequence, from observations. We examine the relation of the resulting approach to other statistical methods for inverse reinforcement learning via analysis and experimental results. We show that preferences can be determined accurately, even if the observed agent's policy is sub-optimal with respect to its own preferences. In that case, significantly improved policies with respect to the agent's preferences are obtained, compared to both other methods and to the performance of the demonstrated policy.