A computer experiment and consideration of rule extraction methods by rough sets based on a rule space

Rough sets are often used for extracting rules from categorical data sets with condition and decision attributes. However, the conventional method used to extract these rules has difficulty in penetrating the extracting processes and in examining the validity of the results. In this paper, we first construct a simulation model, and experimentally review and discuss the conventional method in a rule space which consists of atom rules. We then propose a method that extracts insightful results in the rule space. Through review and discussion, we further propose an effective algorithm for extracting rules, and study the influences on the data arising from inconsistent and unconcerned rules from human judgment and errors by misjudgment.