A novel method for measuring the dissimilarity degree between two pieces of evidence in Dempster-Shafer evidence theory

Abstract

Dempster-Shafer (D-S) evidence theory is one of the most prevalent methods of uncertainty reason in reliability engineering and information fusion. However, the conflict factor cannot fully describe the degree of conflict between two pieces of evidence in classical D-S evidence theory. In this paper, the existing methods in terms of characterizing the dissimilarity degree of evidence are reviewed and a novel dissimilarity measure is proposed based on the Jousselme distance and the conflict factor. The properties of the novel dissimilarity measure, including its non-negativity, symmetry, and boundedness are demonstrated, respectively. To verify the effectiveness of the novel the dissimilarity measure, the classical example is used to compare the performances of the proposed dissimilarity measure with other methods.