The paper combines and extends the technologies of fuzzy sets and association rules, considering users’ differential emphasis on each attribute through fuzzy regions. A fuzzy data mining algorithm is proposed to discovery fuzzy association rules for weighted quantitative data. This is expected to be more realistic and practical than crisp association rules. Discovered rules are expressed in natural language that is more understandable to humans. The paper demonstrates the performance of the proposed approach using a synthetic but realistic dataset.
David L. Olson, Yanhong Li