The Design And Implementation Of Agent Individual Oriented Belief Revision System

For an agent individual, it receives information from external environment or commu-nication with other agents. The more information it gets, the more rational its decision is. But the agent cannot make sure that the new information is consistent with existing beliefs. So it is necessary that belief revision should be operated in order to guarantee belief base’s consistency. Although belief revision has been intensively studied by researchers in the field of artificial intelligence, most of them is researched under full first order logic. Due to the cost of checking consistency in first order logic, the implementations of belief revision sys-tem based on these research results is poor in performance. Otherwise, to some extent, the existing belief revision systems. have defects in function, model and visualization, which hinders the application and development of belief revision.After the analysis of characters in existing belief revision system, we summarize several critical problems to be solved when constructing a belief revision system. In order to solve these problems, this paper proposes a belief revision system model for agent individual, gives the implementation of it and explores the architecture, technology, development and evaluation when constructing the belief revision system.Firstly, the problems existing in current belief revision system are analysed. After s-tudying the existing belief revision theories and their corresponding implementation, for the definition of agent, we conclude that rationality and implementability should be traded of-f in order to seek a balance between them. Therefore, this paper adopts the definition of resource-bounded agent and its corresponding implementation method of belief revision, which which lays a solid foundation for the system’s processing efficiency.Secondly, this paper proposes a model of agent individual oriented belief revision sys-tem. In this model, MongoDB characterized by distributed document-oriented storage and BOSN’S universality in data transmission is used to store belief base, which can improve the expandability of system’s storage and the integrity and operation efficiency of belief.For the sake of better regulation for transmitted message between agents and enhancement of agen-t’s compatibility with web application in data format, this paper integrates ACL’s semantic with JSON’s grammar so as to present a novel agent communication language JSON-ACL.Finally, this paper gives a implementation of belief revision system. According to the model proposed in this paper, we implement the functions of all module by means of Java GUI. Especially, the module of belief base visualization can reveal the integral structure of different status in form of graph, which provides a novel description method for belief revision.The experiment results show that the belief revision System designed and implemented in this paper gets a good performance on the scale of processing problem, interpretability, and visibility.