There are several approaches for evaluating conflict within belief functions. In this paper, we develop one of them based on axioms and show its connections to the decomposition approach. We describe a class of conflict measures satisfying this system of axioms and show that measuring conflict can be realized through the clusterization of a body of evidence. We also show that well-known conflict measures like the auto-conflict measure and the measure of dissonance do not satisfy the proposed system of axioms. We also tackle the problem of simplifying a body of evidence based on clusterization and show the application of developed theoretical constructions for data processing.

有几种方法可以评估信念函数内的冲突。在本文中，我们基于公理开发了其中一个，并展示了它与分解方法的联系。我们描述了一类满足该公理系统的冲突度量，并表明可以通过大量证据的聚类来实现冲突度量。我们还表明，众所周知的冲突测量，如自动冲突测量和不协调测量不满足所提出的公理系统。我们还解决了基于聚类简化证据主体的问题，并展示了已开发的理论结构在数据处理中的应用。