Using action benefits and plan certainties in multiagent problem solving

Abstract

Choosing socially rational actions is essential in multiagent problem solving. In the problem domain of the diagnosis of internetwork problems, exchanging agents' plans is helpful for understanding what are the most rational actions. If they have limited shared knowledge or a shared environment, however, it is hard to understand other agents' plans. The author discusses utility-based cooperation for this situation. A utility matrix is created based on local plans and through communications with other agents, instead of by exchanging plans. Utility numbers are calculated based on action benefits and plan certainties. Intuitively, an action benefit expresses the importance of performing or verifying the current plan, and a plan certainty expresses how strongly the agent making the plan believes that it is correct or effective for the current problem solving. Actions based on a plan supported by many proofs have high utility-numbers and so have priority over other actions. It is shown, through experiments, that performance can be improved by this method. >