The problem of imperfect knowledge has been tackled for a long time by philosophers, logicians and mathematicians. The main idea of rough set theory is to extract decision rules by attribute reduction and value reduction in the premises of keeping the ability of classification, reducing the condition attributes based on the extension set theory and rough set method, calculating the importance to the decision attribute for each condition attribute after reduction, and data mining the relevant rules based on the reduced attributes .To combine extension sets with rough sets pair analysis, a new synthetic evaluation approach is proposed. The result shows this method is quite valuable in knowledge discovery for large category complex information systems.