动态k近邻辅助多权值Slope One算法  

Dynamic k-nearest neighbor-assisted multi-weight Slope One algorithm

作　　者：马浩[1] 黄俊[1] 孔麟 郑小楠 郭小焕 MA Hao;HUANG Jun;KONG Lin;ZHENG Xiao-nan;GUO Xiao-huan(College of Communication and Information Engineering,Chongqing University of Posts and Telecommunications,Chongqing 400065,China)

机构地区：[1]重庆邮电大学通信与信息工程学院,重庆400065

出　　处：《计算机工程与设计》2020年第11期3072-3077,共6页Computer Engineering and Design

基　　金：国家自然科学基金项目(61671095)。

摘　　要：为有效解决传统Slope One算法推荐精度低的问题,提出一种动态k近邻辅助多权值Slope One算法。结合k近邻的思想,对相似度计算方法改进,筛选大于相似度阈值的近邻用户集进行平均评分偏差计算,减少大量干扰评分数据带来的影响。使用用户相似度、用户综合信任度和项目相似度作为权值加权到评分预测当中,进一步提升推荐精度。将算法应用于MovieLens数据集,与几种算法进行对比,实验结果表明,改进的算法均优于其它几种算法,有效提高了推荐的质量。To effectively overcome the problem that the traditional Slope One algorithm has low recommendation accuracy,a dynamic k-nearest neighbor-assisted multi-weight Slope One algorithm was proposed.The idea of k-nearest neighbors was combined and the similarity calculation method was improved.The neighboring user sets larger than the similarity threshold were screened to calculate the average score deviation,and the influence of a large amount of interference score data was added.User similarity,user comprehensive trust,and project similarity were used as weights to weight the score prediction,further improving the recommendation accuracy.The algorithm was applied to the MovieLens dataset and compared with several algorithms.Experimental results show that the improved algorithm is better than several other algorithms,which effectively improves the quality of recommendation.

关 键 词：Slope One算法 K近邻 相似度 多权值 评分预测 

分 类 号：TP391[自动化与计算机技术—计算机应用技术]