A method of evidence of combination based on the improved similarity measure

Aiming to the phenomenon of counter intuition when dealing with some highly conflictive evidences by using the D-S combination rule, the reason for the occurrence of conflict is deeply analyzed. That is to say, it is caused by the outer conflicts among evidences and inner conflicts within the evidence itself. And then, the similarity measure is improved by considering inner and outer conflicts. The credibility is used to improve the corresponding evidential sources according to the improved similarity measure. The original evidences are clustered by using the ISODATA method. Each clustering is combined by using the unified combination rule according to its reliability computed. Finally, an example is given to compare this method in this paper with others and itself before improving. It shows that the new method solves the problem of counter intuition when combining the highly conflictive evidences and has the distinct advantage over others.