诊断系统中不精确推理规则的可靠度及症兆的置信度

本通过对诊断系统中专家知识的不精确性以及症兆信息模糊性的讨论，建立了关于症状重要性比较的语言值判断矩阵，并采用量化的方法将语言值矩阵转化为正互反矩阵，在解此正互反矩阵特征值的基础上，给出了不精确推理规则的可靠度的计算方法，同时还给出了症兆置信度和原因的可能性大小的计算公式，并应用该方法对一个实际问题进行了计算。

Reliability and Symptom Confidence oflnexact Reasoning Rule in Diagnosis

Through the discussion of an inexactness of expert's knowledge and fuzziness of symptom in diagnosis, this paper constructs a languange judgment matrix for the important comparison of symptoms, and converts it into a positive reciprocal matrix by quantified metheod. On the basis of solving eigen value of the positive reciprocal matrix, this paper also gives a computation method for graduating the rule reliability and symptom confidence of inexact reasoning. Finally we present an actual example to demonstrate our technique.