Some methods of handling inconsistency in probabilistic knowledge base

Abstract

To solve the inconsistency of probabilistic knowledge bases, several strategies have been proposed and developed by changing the structure of the components inside probabilistic knowledge bases, removing a part of knowledge, and assessing the degree of inconsistency by using inconsistent measures that change the probability value of knowledge. These strategies typically construct a family of operators to convert an inconsistent probabilistic knowledge base to a consistent one. This paper focuses on the third approach, considering changing the probability values of probabilistic constraint in the probabilistic knowledge base towards a simpler direction. This paper synthesizes two operators and proposes a new one to restore the consistency of the probabilistic knowledge base. A model for dealing with inconsistencies in the probabilistic knowledge base is also provided. A resolution algorithm is proposed for each consistency restoring operator. Additionally, the cost of each algorithm, as well as the desirable principles of operators, are taken into account.