Research On Key Problems Of Agent Development And Interaction

With the rapid development of computer and network technology, people rely more and more on network in communication. Agents and Multi-Agent System (MAS) with characteristics of autonomous and intelligence can improve the efficiency of information processing in communication through network. However considering the internal complexity of modeling and implementation of Agent and MAS, developers need a method and related tools of supporting rapid development of systems based on Agent and MAS. Besides, agents may face some problems hard to process, like information asymmetry and trust establishment. Then agents need to show the characteristic of proactive to dealing with these problems. This paper works on the methods of agent rapid development as well as processing of complex problems communication in a proactive manner, and gets results as follows:This paper illustrate an agent development framework, in which Agent Pattern Language(APL) is used to depict related elements like goals or types of the scenario, and agents with built-in blackboard structure parse APL files and execute as the process specified in APL files. Besides, the paper illustrates the layered MAS structure based on the development framework to support enterprise-oriented application integration. With such a development framework the developers can focus on modeling the scenario with APL without considering how agents work, and agents and MAS can be developed in a rapid way.We also propose a communicating process between agents with a proactive manner to deal with problems of information asymmetry and trust establishment. Such a process is modeled with a state-space equation. The state-space corresponds to the reason process of identifying and processing of input and output messages. On the basis of the state-space equation, the process of dealing with information asymmetry and trust establishment is further introduced.From the cognitive view, this paper introduces a presentation of information asymmetry between agents, as well as the communicating process of dealing with information asymmetry on the basis of the state-space equation. In the communicating process the agent short of information proactively requests information from the agent with the information. As the agent with the information can take advantage of information by hiding information in communication, this paper also presents a way of game on information hiding to constraint the behaviour of two agents that are involved in information asymmetry. Then with proper designed game mechanism in the concrete scenario, information asymmetry can be effectively dealt by the communicating process presented in this paper.Also from the cognitive view, we discuss the definition of trust in the situation of delegation and non-delegation. In view of agents face difficulty to establish trust with game theory based model as lacking of direct experience and witness information, we propose agents establish trust with others by direct communicate with them to collect information with the process depicted with the state-space equation. As the reliability of information in communication should be considered, we introduce a fuzzy inference process on the basis of cognitive reasoning process. With such a manner, the dynamic context changes can be reflected in the trust establishment and trust can be established in a more flexible way.With the development framework proposed here, developers can implement agent and MAS in a rapid way. Besides this paper proposes the proactive communicating process to deal with information asymmetry and problem of trust establishment, which can reduce people's interference in communication between agents and help people to make better use of ability of information process of agents. This paper also provides some referring significance for research of proactive behaviour in MAS.