The generalized combination of evidence involved a priori knowledge

This paper describes a theory called conditional Dempster-Shafer (CDS). This theory provides a unified frame to solve the knowledge updating problem about an uncertain variable. We assume that a priori knowledge about an uncertain variable is modelled by a fuzzy measure and the evidences under the underlying uncertain variable are modelled by the Dempster-Shafer belief measures. The updated knowledge about the uncertain variable is still represented by a fuzzy measure. The updating process of knowledge is as follows: the evidences about the uncertain variable are firstly fused by the rule of conditioned Dempster-Shafer combination proposed in this paper, then the updated knowledge about the variable is achieved using the conditioned agreement function on the combined evidence. We show that the updated knowledge can be represented as a conditional fuzzy measure given the evidences.