Learning User Preferences for Multi-attribute Negotiation: An Evolutionary Approach

This paper investigates how agents that act on behalf of users in electronic negotiations can elicit the required information about their users’ preference structures. Based on a multi-attribute utility theoretic model of user preferences, we propose an algorithm that enables an agent to learn the utility function over time, taking knowledge gathered about the user into account. The method combines an evolutionary learning with the application of external knowledge and local search. The algorithm learns a complete multi-attribute utility function, consisting of the attribute weights and the individual attribute utility functions. Empirical tests show that the algorithm provides a good learning performance.