Analyzing the combination of conflicting belief functions

作者: Philippe Smets

DOI: 10.1016/J.INFFUS.2006.04.003

关键词: Empty setTransferable belief modelDynamic dataMathematicsBelief structureMathematical economicsUncertain dataArtificial intelligenceNormalization (statistics)

摘要: We consider uncertain data which uncertainty is represented by belief functions and that must be combined. The result of the combination can partially conflictual. Initially Shafer proposed Dempster's rule where conflict reallocated proportionally among other masses. Then Zadeh presented an example produces unsatisfactory results. Several solutions were proposed: TBM solution masses are not renormalized stored in mass given to empty set, Yager's transferred universe Dubois Prade's resulting from pairs conflictual focal elements union these subsets. Many suggestions have then been made, creating a 'jungle' rules. discuss nature combinations (conjunctive versus disjunctive, revision updating, static dynamic fusion), argue about need for normalization, examine possible origins conflicts, determine if justified analyze many solutions.