DATA LATENT RULE EXTRACTION DEVICE, DATA LATENT RULE EXTRACTION METHOD AND DATA LATENT RULE EXTRACTION PROGRAM

摘要

PROBLEM TO BE SOLVED: To provide a device for data classification and data latent rule extraction by a fuzzy method. SOLUTION: This data latent rule extraction device 1 has a means 10 for inputting individual data units from a data set; a means 20 for eliminating the data unit which is off a prescribed reference; a means 30 for normalizing each item element of each data unit which is not eliminated, according to commonalization rules to make the data unit a multidimensional fuzzy vector; a means 40 for computing normal distribution on an element of the multidimensional fuzzy vector and uniting the normal distribution of the data units whose normal distribution are similar, to generate categories for classifying the data units; and a means 50 for generating an inference rule group by extracting a condition part and a conclusion part of each knowledge expression element from the dimensions of the data units classified in categories. COPYRIGHT: (C)2004,JPO