摘要
协同过滤技术是目前应用最多的一种推荐技术,这项技术从用户提供的信息中展开发掘,按"物以类聚,人以群分"的原则产生和目标用户(或项目)相似性高的最近邻,从中预测评分,进而产生推荐。但是由于评分信息稀疏化就会造成无法适应用户兴趣,而且推荐的实时性差等问题。针对上述问题,文章提出了一种带有改进的用户-项目类型喜好相似性的计算方法完善用户兴趣改变的问题,并且结合了优化后的双重k-means聚类,使搜索最近邻的范围大大减少,从而提高了推荐算法的实时性。实验结果表明,该优化后的协同过滤推荐算法能通过时间相似性更好地适应用户兴趣的变化,推荐的精度最精确,效果更易使用户满意。
Collaborative ifltering technology is a kind of recommendation technologies currently used the most, discovery of this technology from the information provided by the user, according to the ＂Like attracts like., produce Birds of a feather lfock together.＂ principle and target users （or items） nearest neighbor similarity is high, a predictive score from birth, and then recommend. But the score information sparse will result can not adapt to the user interest, problems and poor real-time recommendation. Aiming at the above problems, this paper proposes a with an improved user item type preferences similarity calculation method to improve user interest change problem, and combined with the dual K-means clustering after optimization, the scope of the nearest neighbor search is greatly reduced, thus improving the real-time performance of the recommendation algorithm. The experimental results show that, collaborative ifltering recommendation algorithm of the optimized through time similarity to better adapt to the changes of the user＇s interests, the most accurate recommendation accuracy, easy to use user satisfaction effect.
出处
《无线互联科技》
2015年第5期124-127,共4页
Wireless Internet Technology
关键词
协同过滤
用户兴趣
双重聚类
K-means优化
调和相似性计算
Collaborative ifltering
User interest
Double cluster
K-means optimization
Harmonic similarity calculation
