Cooperative Multiagent Framework for Knowledge Retrieval
Florin Leon, Mihai Horia · 2003
In this paper two methods for improving the intelligent agents framework base structure are presented. Those are very important from the practical implementations point of view because there is a great distance between the formal description of problem solving and the modifications that need to be made for obtaining the best fit with an existing general support frame. In this paper two specific problems are analyzed. The first one is typical for agents because they have their own autonomy and is related to the inter-agent communication problem. The second is related to the previous one but uses a classical approach. Agent communication is needed in order to quickly increase their knowledge based. This is important because they serve dierent entities or they have dierent tasks to accomplish. Although we must take into account that the framework must not generate overload over the hosting systems both at the level of used computing power and to network bandwidth. In order to do that a new method for communication decrease is proposed. The main idea resides in using a cooperation factor to indicate the probability that an agent can answer a question from its peers. The second approach tries to solve the priority problem. This will also decrease the