A Generic Rule for Combination of Uncertain Information in the Framework of Evidence Theory

The body of work known as evidence theory has attracted much attention in dealing with imperfect and uncertain information during the past three decades. Many different formal techniques have been developed in order to address the method of information aggregation and conflict handling; however none of the models available today can fit with all kinds of imperfection and uncertainty. The models based on mass multiplication assume conditional independence of information sources, but unfortunately these assumptions are far from realistic in many real world situations. In this paper we propose a unifying formalism by merging the generalized fuzzy set operations based on triangular norms, with belief functions computation. This generic framework is not only capable of retrieving other evidential models’ combination methods, but also enables us to define a new family of combination rules which can deal with a complete range of dependency assumptions.