青岛大学学报（自然科学版）2018，Vol.31Issue(2)：116-120,5.DOI:10.3969/j.issn.1006-1037.2018.05.20
基于用户协同过滤的近邻选择算法
A Neighbor Selection Algorithm Based on Users' Collaborative Filtering
摘要
Abstract
Collaborative filtering recommendation algorithm based on user is mainly according to user's neighbor ratings to predict the score of unknown item.In neighbor-selecting phase,traditional methods need to manually set threshold of common ratings or user similarity as a filter criterion,which could ensure the credibility of neighbor selection.But the threshold is not easy to set up in reality.It was made a formal description for neighbor-selecting standards based on Hoeffding boundary theory,and could be adaptively selected neighborhood according to user's co-ratings.The experiment proved that compared with traditional methods,this method can increase the reliability of the user neighbor,and enhance prediction accuracy.关键词
推荐系统/近邻选择/协同过滤Key words
recommendation system/neighborhood-selection/collaborative filtering分类
信息技术与安全科学引用本文复制引用
孙兰兰..基于用户协同过滤的近邻选择算法[J].青岛大学学报（自然科学版）,2018,31(2):116-120,5.基金项目
安徽省精品开放课程视频公开课(批准号:2016gkk033)资助 （批准号:2016gkk033）
桐城师范高等专科学校自然科学研究重点项目(批准号:TCSZZR2017B01)资助. （批准号:TCSZZR2017B01）
