A new uncertainty measure for belief networks with applications to optimalevidential inferencing

This paper is concerned with the problem of measuring the uncertainty in a broad class of belief networks, as encountered in evidential reasoning appl ications. In our discussion, we give an explicit account of the networks co ncerned, and coin them the Dempster-Shafer (D-S) belief networks. We examin e the essence and the requirement of such an uncertainty measure based on w ell-defined discrete event dynamical systems concepts. Furthermore, we exte nd the notion of entropy for the D-S belief networks in order to obtain an improved optimal dynamical observer. The significance and generality of the proposed dynamical observer of measuring uncertainty for the D-S belief ne tworks lie in that it can serve as a performance estimator as well as a fee dback for improving both the efficiency and the quality of the D-S belief n etwork-based evidential inferencing. We demonstrate, with Monte Carte simul ation, the implementation and the effectiveness of the proposed dynamical o bserver in solving the problem of evidential inferencing with optimal evide nce node selection.