【摘要】 <正> 1.前言在学习系统或知识库系统等智能应用中,都要解决如何将新知识综合到系统原有的信念集中。最初的研究方法采用真值维护(TMS,Truth Maintenance System),但只能在孤立上下文中起作用。八十年代中期,G(?)rdenfors、Alchourron与Makinson等人共同创立了信念修正理论(BR,Belief revision),解决了开放上下文中的知识维护,后继研究者通常将更多还原

【Abstract】 Belief Revision is a theory that studies how to integrate new information into original belief set. Classical BR theory uses AGM frame, but it only resolves problems in single agent BR system. Multi-agent BR faces problems such as the collision of many information sources and how to maximize the logic consistence of multi-agent system. On the basis of game theory model, we form profit matrix under different BR strategies in multi-agent system and try to get the best strategy that satisfies logic consistence of the system through negotiation.更多还原