摘要
针对传统推荐算法的数据稀疏性问题和推荐准确性问题,提出基于粒子群优化的项聚类推荐算法。采用粒子群优化算法产生聚类中心,在此基础上搜索目标项目的最近邻居,并产生推荐,从而提高了传统聚类算法的推荐准确性及响应速度。实验表明改进的项聚类协同过滤算法能有效提高推荐精度。
Aiming at the problems that the data are sparse and the results are not accurate in traditional recommendation algorithms, this paper proposes an item clustering recommendation algorithm based on Particle Swarm Optimization（PSO） algorithm. It uses PSO to engender the cluster centers, calculates the similarity between target item and cluster centers to search the nearest neighbors of target item, and gains a recommendation, so that it improves the accuracy and the real-time performance. Experimental results indicate that the algorithm can effectively improve the accuracy of the recommendation system.
出处
《计算机工程》
CAS
CSCD
北大核心
2009年第23期178-180,共3页
Computer Engineering
基金
教育部留学回国人员启动基金资助项目(教外司留[2007]1108-10)
关键词
粒子群优化
项聚类
协同过滤
推荐算法
Particle Swarm Optimization（PSO）
item clustering
collaborative filtering
recommendation algorithm
