The Design And Implementation Of A Self-Organizing Concurrent Negotiation Model

The rapid growth of the computer techniques makes more and more people rely on computer in their daily works, or even delegate them to intelligent agents. When a problem is encountered, an agent may not have sufficient knowledge, therefore need to find others with relevant expertise for help.Intelligent agents need relevant expertise to effectively achieve automatic negotiation. Search is an effective approach to find relevant expertise. However, most search algorithms available nowadays assume that the target space be specified in a clear and explicit manner. They have difficulties when only obscure, incomplete and unstructured information is provided. The huge success in search engines domain suggests that information retrieval is an effective alternative to deal with such cases.This paper presents a concurrent negotiation method which is based on information retrieval technology. In this model, the agent analyzes other agents' expertise, and based on its own need, identifies those with relevant expertise, before grouping them into clusters. Each member of the cluster can then join the concurrent negotiation, before a consensus (i.e. solution to the problem encountered) is reached via information exchange. This negotiation process can be limited within a certain number of negotiation rounds. Agent can learn the others problem-solving abilities from this process. Such knowledge could be shared among the agents, which could further enhance their problem-solving abilities.