An extended influence diagram (EID) is a graph model for modelling and solving decision problems under uncertainty, extended from the influence diagram. Because statistical data is not enough, available data may not be directly use in the network, domain experts may be reluctant to provide accurate probability, probability elicitation is still a major obstacle for the developing of probabilistic network. In this paper, we propose a bottom-up approach for probability elicitation based on the structure of EID. This approach combines the qualitative and quantitative information, mainly uses interval probability for analyzing. It is useful for reducing the experts' burden and avoiding the bias problem among experts.
展开▼
