Learning on opponent's preferences to make effective multi-issue negotiation trade-offs

Software agents that autonomously act and interact to achieve their design objectives are increasingly being developed for a range of e-commerce applications. In this context, automated negotiation is a central concern since it is the de facto means of establishing contracts for goods or services between the agents. Now, in many cases these contracts consist of multiple issues (e.g. price, time of delivery, quantity, quality) which makes the negotiation more complex than when dealing with just price. In particular, effective and efficient multi-issue negotiation requires an agent to have some indication of its opponent's preferences over these issues. However, in competitive domains, such as e-commerce, an agent will not reveal this information and so the best that can be achieved is to learn some approximation of it through the negotiation exchanges. To this end, we explore and evaluate the use of kernel density estimation for this purpose. Specifically, we couch our work in the context of making negotiation trade-offs and show how our approach can make the negotiation outcome more efficient for both participants.