Specifying the Processes and States of Negotiation

Negotiation can be considered to be an important aspect of commerce. It is part of those wider dynamic processes whereby commercial goals are achieved by the parties to a contract. Overt negotiation, as deal making, is often suppressed by agreed rules of encounter, but it is rarely absent altogether. In this paper, while recognising the need for more complete logics to represent both states and processes with abstraction, we start to build a more formal link between negotiation and Artificial Intelligence. We illustrate the use of dynamic logic to specify a shopping scenario between a retailer and a customer agent. The negotiation that arises in such a scenario can follow one of the negotiation models described. From a given negotiation model we obtain the corresponding negotiation protocol. These formulations have allowed us to remove inconsistencies and ambiguities in less formal models and to suppress issues such as concurrency. Finally, we discuss how an agent, given that it has mental states and a library of plans, can find a path for negotiation that will not only be successful in achieving its goal but also be optimal.