On a Method to Extract Rules from a Table with Nondeterministic Information: a Rough Sets Based Approach

Rough sets theory is now becoming a mathematical foundation of soft comput ing. This theory makes use of equivalence relations defined for each set of attributes in any table, and applies the concept like definability of a set, dependency among attributes, reduction of data, rule extraction, etc., to data analysis. In this paper, a problem of knowledge discovering in the form of rules from any table with nondeterministic information is discussed. At first, the rough sets based concept including rule extraction is surveyed, and this concept is extended to new one related to nondeterministic information. Then, a framework of rule extraction from tables with nondeterministic information is proposed, and some algorithms for handling such new concept are presented. Also implemented programs and a real execution of these programs are shown.