Searching Decision Rules in Very Large Databases Using Rough Set Theory

Abstract

This paper applies rough set theory to find decision rules using ORACLE RDBMS. The major steps include elimination of redundant attributes and of redundant attribute values for each tuple. In this paper three algorithms to extract high frequency decision rules from very large decision tables are presented. One algorithm uses pure SQL syntax. The other two use sorting algorithm and binary tree data structure respectively. The performances among these methods are evaluated and compared. The use of binary tree structure improves the computational time tremendously. So pure SQL results indicate some major change to query optimizer may be desirable if it will be used for data mnining.