【摘要】 Ｄ．Ｄｕｂｏｉｓ和Ｈ．Ｐｒａｄｅ提出的可能性逻辑是一种基于可能性理论的非经典逻辑，主要用于不确定证据推理。可能性逻辑不同于模糊逻辑，因为模糊逻辑处理非布尔公式，其命题中包含模糊谓词，而可能性逻辑处理布尔公式（最简单的情形），其中只包含经典命题和谓词。本文尝试在可能性理论的框架下进行不相容知识库的维护和问题求解。这里的知识表示是基于可能性逻辑的。为此，我们提出了两种不同的方法：第一种方法在计算命题（关于不相容知识库的）可信度时，要考虑所有极大相容子集；第二种方法是基于信念维护的，它通过选择最佳极大相容子集来恢复知识库的一致性。更多还原

【Abstract】 Possibilistic logic proposed by D. Dubois and H. Prade, is a possibility theorybased non--classical logic for reasoning under uncertain evidence. The difference of possibilistic logic from fuzzy logic is that, while fuzzy logic manipulates a set of non-Boolean formulas involving vague predicates, possibilistic logic (under its simplest form) manipulates a set of Boolean formulas in which involved classical propositions and predicates only. In this paper, we try to deal with the revision and the question answering of inconsistent knowledge bases in the framework of possibility theory. Here the knowledge is represented in the light of possibilistic logic. Two different approaches has been presented for this purpose. The first way takes into account all the maximal consistent subsets when evaluating the degree of confidence of a proposition with respect to an inconsistent knowledge base; the second one is a belief revision-based approach, which consists in restoring the consistency by selecting the optimal maximal consistent subsets.更多还原