Researching On Some Uncertainty Reasoning Methods

Uncertainty reasoning is one of focus research domain in Artificial Intelligence. At present, some uncertainty reasoning methods used frequently such that probability reasoning, evidence reasoning , vague reasoning , rough set reasoning, each of these methods which exists in uncertainty reasoning methods has its advantages and disadvantages and has been always developed . In this paper evidence theory, rough set theory and probability reasoning method have been discussed.Within the evidence reasoning, data fusion consists in obtaining a single belief function by the combination of several belief functions resulting from distinct information sources. The most popular rule of combination is Dempster's rule of combination. And then some outstanding new models have been given. The merits and demerits of these models are studied, especially the Transferable Belief Model has been studied in this paper.In rough set theory, the accuracy measure is an important numerical characterization that quantifies the imprecision of a rough set caused by its boundary region. However, the traditional accuracy measure does not take into consideration the granul arity of the partition induced by an equivalence relation. This paper first analyzes the limitation of the traditional accuracy measure and the one proposed by Baowen Xu etc. and then proposes a new kind of definition of accuracy measure . And some good properties of it have been proved, What's more, it is more suitable to measure the imprecision of rough sets and its algorithm is simpler than Baowen xu etc's which are illustrated by examples.A Bayesian Belief Network (BBN) is a graphic model that encodes joint probability distribution among uncertain variables, it express a potential dependent relationship between variables. Modeling with Bayesian belief network has been a powerful tool to solve many uncertainty problems. This paper proposes a new approach for computing probabilities of events in Bayesian network. It makes probability reasoning of events easier and the results makes us satisfied.