Prediction of partners' behaviors in agent negotiation under open and dynamic environments

Prediction of partners' behaviors in negotiation has been an active research direction in recent years in the area of multi-agent and agent system. So by employing the prediction results, agents can modify their own negotiation strategies in order to achieve an agreement much quicker or to look after much higher benefits. Even though some of prediction strategies have been proposed by researchers, most of them are based on machine learning mechanisms which require a training process in advance. However, in most circumstances, the machine learning approaches might not work well for some kinds of agents whose behaviors are excluded in the training data. In order to address this issue, we propose three regression functions to predict agents' behaviors in this paper, which are linear, power and quadratic regression functions. The experimental results illustrate that the proposed functions can estimate partners' potential behaviors successfully and efficiently in different circumstances.