Investigating adaptive, confidence-based strategic negotiations in complex multiagent environments

We have designed and implemented an adaptive, confidence-based negotiation strategy for conducting multiple, concurrent negotiations among agents in dynamic, uncertain, and real-time environments. Our strategy deals with how to assign multiple issues to a set of concurrent negotiations. When an agent is confident about a particular peer agent, it uses a packaged approach by negotiating on multiple issues with that peer in a single negotiation job. Otherwise, it uses a pipelined approach by negotiating with the peer one issue at a time in a sequence of negotiation jobs. Hence, the confidence of an agent’s profile or view of other agents is crucial, and that depends on the environment in which the agents operate. Each initiating agent is also motivated to improve both the process and the outcome of its negotiations, taking into account factors such as time spent and messages sent during negotiation. Our experiments show that the adaptive, confidence-based negotiation strategy outperforms the purely pipelined or purely packaged strategy in a variety of aspects.