Association rule mining of Kansei knowledge using rough set

In the field of Kansei Engineering, Semantic Differential (SD) is a typical method in evaluating conceptual products. We design a WEB-based kansei questionnaire system to generate a Decision Table(DT) which is composed of typical form features (condition attributes) and kansei words(decision attributes) through SD methods; by using statistical tools, frequent records are stored to DT as decision rules which are indexed by kansei word. First, some rules which have important contributions to the corresponding kansei evaluation will be extracted through the attribute reduction algorithm of Rough Set Theory(RST). Second, the size of decision table has been reduced through association rule mining based on Rough set and rules joining operating; finally, strong association rule set which describe the relation between the key form feature and the corresponding kansei word is generated. The proposed method has been successfully implemented in cell phone design case.