Coping with conflict in cooperative knowledge-based systems

In this paper, we address a critical issue of cooperative problem solving: the existence of conflict among distributed agents. In particular, we focus our study on cooperative knowledge-based systems. To obtain a better understanding and more balanced judgement of multiagent conflict, we provide a general scheme to study the logical structure of multiagent conflict and rational strategies of coping with it under different situations. Our research finding is that there is no grand unified theory of coping with conflict in performing complex real-world computer supported tasks. Instead, a library of alternative methods should be considered. We discuss four methods: inquiry, arbitration, persuasion, and accommodation. These methods can be combined in an order appropriate to the application domain such that if one method fails, the system will try the next. We point out merits and shortcomings of these methods and illustrate them using several high-level protocols and application examples from a prototype system, the Building Design Network