Abstract: Applying rule schemas to interactive association rule mining can improve the participation of users effectively;however if the schemas are larger,much effort has to be spend on rule schemas and their levels.LRU is introduced to optimize rule schemas,and IAFBLRU is proposed in this paper,in which users can proactively improve high interesting rule schemas' level;and also LRU policy can reduce the uninterested rule schemas' level automatically.Experiments show that the algorithm improves the quality of association rule mining algorithm effectively.