改进的FP-Growth算法在个性化推荐中的应用研究

Research and Application of Improved FP-Growth Algorithm In Personalized Recommendation

文章研究了聚类算法K-means和关联规则算法FP-Growth,针对两种算法存在的不足,提出了对原始事务数据库采用K-means凝聚层次算法按购物用户的属性特征进行聚类,并针对聚类压缩后事务数据库,利用矩阵存储数据并按支持度计数排序,并将其应用于在线购物系统的个性化推荐中。算法的提出压缩了事务数据库,降低了生成频繁1项集的时间,减少了FP-Growth算法扫描数据库的次数,提高了算法的执行效率。

This paper studies the clustering algorithm K-means and association rule algorithm FP-Growth.In view of the shortcomings of the two algorithms,it is proposed to cluster the original transaction database by AGglomerative NESting algorithm according to the attribute characteristics of shopping users.For the transaction database after clustering compression,the data is stored by matrix and sorted by support count,it is applied to the personalized recommendation of online shopping system.The proposed algorithm compresses the transaction database,reduces the time to generate a frequent items,reduces the number of times FP-Growth algorithm scans the database,and improves the execution effi ciency of the algorithm.