Here, dimensions (x, y) of the vector space represent words used in judging whether information is inappropriate information or not, and coefficients of the linear discrimination function (such as a and b for a straight line given by ax+by +c=0) are given by the weights of the respective words.