This article proposes an automatic generation method of relation matrix for Interpretive Structural Modelling (ISM), which is a popular tool widely used for visualising problem structure, problem structure analysis, and decision making. In ISM, problems are divided into sub problems. Then hierarchy relations are introduced into the set of sub problems. The hierarchy structure is represented by a directed graph (ISM diagram). Users can clearly understand the hidden problem structures by browsing the ISM diagram.
Drawing an ISM diagram is time-consuming. First, keywords related to the problem are selected, and then the relation between an arbitrary pair of two keywords is identified. If there is a relation between the selected two keywords, the existence of the relation is confirmed. After examining all pairs of any two keywords, all relations are represented as an n x n matrix called a relation matrix. Next a reachable matrix is calculated from the relation matrix. By using the reachable matrix, a hierarchy structure is drawn. After fixing the relation matrix, all calculation will automatically be performed. But fixing the relation matrix, especially identifying the relation between any two keywords, requires time and labour.
In this article, we propose an automatic generation method of the relation between two keywords. First keywords are selected. Then original materials such as newspaper sources are scanned to extract articles that include at least one keyword. Then the number of articles that include any two keywords is counted. Next the co-occurrence probability is calculated, and all probabilities are arranged in a matrix form. Then the average value of co-occurrence probability is calculated for each keyword, and the relation is evaluated based on comparisons of average probability and co-occurrence probability. It is assumed that if co-occurrence probability is larger than average probability, these two keywords have a relatively strong relation.