A cooperative planning algorithm to improve performance in Web domains

Abstract

In this paper, we present MAPWEB, a multiagent framework that integrates planning agents and Web information retrieval agents. The goal of this framework is to deal with problems that require planning with information to be gathered from the Web. Because of flexibility and efficiency reasons, MAPWEB decouples planning from information gathering by splitting a planning problem into two parts: solving an abstract problem and validating and completing abstract solutions by means of information gathering. We focus on the planning process in order to improve its efficiency. There are two ways of improving the efficiency of the MAPWEB planning algorithm: by accelerating the planning process itself and by storing previously solved planning problems. The first has been achieved by designing a cooperative planning algorithm that allows a set of planning agents to share plans and cooperate while planning to increase efficiency. The second is currently being designed and developed to allow planning agents to reuse the acquired knowledge. Finally, this paper presents experimental evaluation of the cooperative planning process when it is used by the planning agents.