Abstract: Bayesian network is the connection of probability theory and graphical model and is widely used to solve the uncertainty problem.But it has deficiencies in dealing with the imprecise information.Possibilistic network is the combination of probability theory and possibilistic theory with graphical model which can remedy defect of Bayesian network.In this paper the concept of possibilistic networks is introduced.Then it is compared with Bayesian network and a method of the possibilistic network structure learning based on the dependency analysis is given.

Key words: possibilistic network, Bayesian network, uncertain information, imprecise information, set-valued

摘要： 贝叶斯网络是概率理论与图形模式的结合，被广泛用于不确定性问题求解，但不具有处理不准确性信息的能力。可能网络是可能理论、概率理论与图形模式的结合，可弥补贝叶斯网络这方面的不足。首先介绍关于可能网络的一些概念，并与贝叶斯网进行比较，然后给出一种基于依赖分析的可能网络结构学习方法。

关键词: 可能网络, 贝叶斯网络, 不确定性信息, 不准确性信息, 集值