No label defined (Q4023482)

The book is a continuation of [Evidence Theory and its Applications, vol. 1; see the preceding review Zbl 0791.68133]. Evidence theory, known in \(AI\) as Dempster-Shafer theory, is a kind of generalization of ordinary probability theory. In evidence theory, basic probabilities are assigned to subsets of the frame of discernment. In probability theory, such assignment is restricted to singletons. Obviously, the worst case time computational complexity of dealing with the possible subsets of the frame of discernment is exponential. Therefore, some techniques for special cases were developed, see e.g. \textit{J. A. Burnett} [Computational methods for a mathematical theory of evidence, Proc. of the 7th IJCAI, 868-875 (1981)] and \textit{G. Shafer} and \textit{R. Logan} [Artif. Intell. 33, 271-298 (1987; Zbl 0633.68093)]. The first five chapters of the book, i.e., Chapters 11-15, present the class of support functions together with discussion of refining/coarsening of evidence. More specifically, inner and outer coarsening mappings, support functions, commutativity of the coarsening operation, commutativity of the refining operation, and commutativity of the orthogonal are studied. The second part of the book, Chapters 16-22, introduces Shafer-Logan linear time computation technique that is applied to a tree hierarchy. The Shafer-Logan's computation technique for hierarchical evidence together with the detailed description of how to implement the computation of the orthogonal sum over a tree hierarchy are presented here.