The paper includes two parts: First, basic theory. Tolerate relation, equivalence relation and unsymmetrical resemble relation were introduced based on rough set theory, upper approximate and lower approximate were defined in incomplete information system. The relationships, three kinds of upper approximate and lower approximate, attribution reduction in incomplete system and core value in complete subsystem, were discussed .Second, algorithm set. 1. A knowledge discovery algorithm of imperfect information based on distinct matrix and decision rules independence principle; 2. A core value ROUSTIDA algorithm based on rough set theory; 3. A dynamic conversion algorithm based on dependence among attributes and its importance.
|