Inference in possibilistic hypergraphs

摘要: In order to obivate soundness problems in the local treatment of uncertainty knowledge-based systems, it has been recently proposed represent dependencies by means hypergraphs and Markov trees. It shown that a unified algorithmic uncertainties via propagation is possible on such structures, both for belief functions Bayesian probabilities, while preserving completeness obtained results. This paper points out same analysis applies approximate reasoning based possibility theory, discusses usefulness idempotence property combining distributions, not satisfied probabilistic reasoning. The second part analyzes previously technique handling dependencies, relating hypergraph approach.