The alpha-junctions are the associative, commutative and linear operators for belief functions with a neutral element. This family of rules includes as particular cases the unnormalized Dempster's rule and the disjunctive rule. Until now, the alpha-junctions suffered from two main limitations. First, they did not have an interpretation in the general case. Second, it was difficult to compute a combination by an alpha-junction. In this paper, an interpretation for these rules is proposed. It is shown that the alpha-junctions correspond to a particular form of knowledge about the truthfulness of the sources providing the belief functions to be combined. Simple means to compute a combination by an alpha-junction are also laid bare. These means are based on generalizations of mechanisms that exist to compute the combination by the unnormalized Dempster's rule.
Keywords. Transferable Belief Model, Dempster-Shafer Theory, Belief Functions, Information Fusion, Uncertain reasoning.
The paper is availabe in the following formats:
Plenary talk : Press here to get the file of the presentation.
Thales Research & Technology
1 avenue Augustin Fresnel
91767 Palaiseau cedex France
U.M.R. C.N.R.S. 6599 Heudiasyc,
Centre de Recherches de Royallieu
F-60205 Compiègne Cedex