A Study on Group Decision Making Problem Based on Fuzzy Reasoning and Bayesian Networks

Aiming at the group decision-making problem with multi-objective attributes, this study proposes a group decision-making system that integrates fuzzy inference and Bayesian network. Based on the mixture of threshold, affiliation function, expert experience and relevant potential background to construct a fuzzy rule base to solve the quantitative problems such as scale differences and expert linguistic variables; designing a hierarchical Bayesian network, constructing a directed acyclic graph with the expert selection as the node, and utilizing the maximum likelihood estimation to dynamically optimize the conditional probability table, resolving the nonlinear correlation between the multidimensional indexes to sum up the a posteriori. This study compares our method with the traditional weighted scoring approach in a comprehensive student evaluation decision-making problem, and the results show that our method performs well in explaining the decision-making in both the construction of rule criteria and the standard ranking scheme. In addition, the performance and robustness of the present method is investigated through computational experiments involving real datasets in the context of different group decision problems.