定性贝叶斯网使用定性的影响标记表示变量之间的影响关系,但不能表达影响力的强弱,在推理过程中容易出现不确定的结果。研究了带权重的定性贝叶斯网,权重是一个[0,1]区间内的数值,用以表达影响力的强弱,修正了推理中的运算规则,并设计了权重标杆方便专家给出权重。保持了定性贝叶斯网络建模过程的简易性,同时提高了推理精度.
Qualitative signs are used in qualitative Bayesian network to express various influences among variables, but the strengths are not indicated. In this paper, we study on qualitative Bayesian networks with weight. A weight is a numerical value in[0, 1] to show the strength of influence. The operation rules used in network reasoning are revised. And we design a weight pole, which can help experts educe weights. This method keeps the simplicity of modeling process and improves the precision of reasoning.