Construction of Decision-Making System for Preventing Internal Control-Type Fraud Using Duality between Positive and Negative Evaluations

处理社会系统或人类群体中的决策问题的重要问题是：（1）如何根据相似性对各种意见进行聚合;（2）如何确定聚合意见的优先级。本文首先利用S型函数表示向量相似度，然后利用模糊关系矩阵的特点，提出了一种新的无矛盾意见聚集方法。在此基础上，提出了一种利用模糊偏好矩阵确定偏好顺序的方法。个人欺诈行为所产生的社会风险取决于（a）欺诈行为的类别、（B）欺诈行为的规模、（c）防范体系的脆弱程度，并由多变量函数表示。对此，我们定义了正反评价的公共参数，并提出了两种不同类型评价的统一方法。通过这些方法，可以提取出社会欺诈易发生的条件。

The important things of dealing with decision-making problem in social systems or human groups are, (1) how to aggregate versatile opinions according to their similarity, and (2) how to find out priority for the aggregated opinions. In thisresearch work, we first exploit sigmoid function method to present vector similarity, and then, applying the characteristics of fuzzy relation matrix, we propose a new opinion aggregating method without contradictions. Relating to these techniques, a method of determining preference order through fuzzy preference matrix is also proposed. The social risk arising from person’s fraud depends on (a) the class of fraud, (b) the magnitude of fraud, or (c) the fragile of defending system, and is presented by multi-variable functions. Concerning this, we define common parameter for positive and negative evaluations and propose the uniting method for the two different type evaluations. Through these methods, we can extract the condition that the social fraud easily occurs.