The strong development of artificial intelligence technology has laid a solid foundation for the construction of litigation rules knowledge base. Currently, in the process of litigation risk analysis, intelligent reasoning has some disadvantages, such as lack of knowledge base of judicial risk rules, low technology of litigation risk analysis, etc., which cannot fundamentally resolve the risks in evidence, limitation of action and code of conduct. Therefore, it is necessary to select cases randomly after they are typed, and collect the indictment, evidence, facts and judgment documents of cases. On the basis of analyzing the multi-party evidence association analysis model, according to the legal risk points, we design and develop the knowledge base of litigation prescription rules, the knowledge base of Party behavior normative rules and the knowledge base of evidence validity rules, combined with multi-party evidence relationship. The joint model, which is integrated with the knowledge base of laws and regulations and the knowledge base of litigation risk rules, uses decision tree algorithm and relational network reasoning technology to enumerate and analyze the possible litigation risks, and finally achieves the identification of litigation risk points, so as to provide comprehensive litigation decision-making guidance, accurate prediction of judgment results and reasonable diversion of unnecessary litigation for the public.