Using Expert's Knowledge to Build Bayesian Networks

Building Bayesian networks is considered to be difficult and time-consuming work. Researchers usually learn networks from data. However, it is of low efficiency because of the huge search space. Since Bayesian networks represent the causal relationships among random variables, domain experts can build Bayesian networks according to their knowledge and experience. In this paper, we develop a method to build Bayesian networks from experts. We use some knowledge elicitation tools to obtain high-quality knowledge, and ensure the validity of networks by combing knowledge from multiple experts. The experimental result indicates that this method can improve the modeling efficiency of Bayesian networks.