贝叶斯网络在因果图中的应用

The Application of Bayesian Network in Causal Map

本文的目的在于描述一种用于表征和分析专家知识“贝叶斯因果图”。它既是因果图 ,也是贝叶斯网络 ,综合了两者在表达专家知识方面的优点 ,并采用事实传递算法对决策变量进行推断。本文以一个市场专家进行产品发展决策为例 ,阐述了如何从因果图构造贝叶斯因果图以及如何用它来进行推断。

The main goal of this paper is to describe a ‘Baye sian Causal maps’ to represent and analyze domain knowledge of experts. Both a c ausal map and a bayesian network. A Bayesian causal map integrates their advanta ges in expressing knowledge of experts, and uses the evidence propagation algori thms to make inference for decision variables. By describing a marketing expert decision process on product development, the paper illustrates how the causal ma ps can be modified to construct Bayesian causal maps and how to make reference.