摘要
针对在传统协同过滤算法中存在的推荐精度较低、预测质量不佳的问题,该文提出一种基于可信预测值的协同过滤算法(RPCF).该算法在使用基于记忆的协同过滤方法计算预测值的基础上,引入可信度概念和技术方法,运用对推荐项目评级的邻居数评估可信度,融合可信度与传统预测值得到可信预测值,再根据可信预测值进行推荐,从而达到提升算法质量的目标.在MovieLens数据集中与其他提高精度方法进行实验对比,实验结果表明:RPCF方法能够提高预测精度和算法鲁棒性,具有更好的推荐质量.
To address the problem of low recommendation accuracy and poor prediction quality in traditional collaborative filtering algorithms,the collaborative filtering algorithm based on reliable prediction value(RPCF)is proposed.On the basis of the predicted value calculated by memory-based collaborative filtering method,the algorithm introduces the concept and technical method of credibility,and employs the number of rated neighbors to evaluate credibility,the credibility and traditional predicted value is fused to obtain the credible predicted value,then the recommendation is generated according to credible prediction value,so as to achieve the goal of improving the quality of the algorithm.Compared with other methods to improve accuracy on the MovieLens dataset,experimental results show that RPCF can improve prediction accuracy and algorithm robustness,and has better recommendation quality.
作者
邓泓
吴祎
于程远
袁徽鹏
DENG Hong;WU Yi;YU Chengyuan;YUAN Huipeng(Software College,Jiangxi Agricultural University,Nanchang Jiangxi 330045,China;School of Computer and Information Engineering,Jiangxi Agricultural University,Nanchang Jiangxi 330049,China)
出处
《江西师范大学学报（自然科学版）》
CAS
北大核心
2022年第6期642-648,共7页
Journal of Jiangxi Normal University(Natural Science Edition)
基金
国家自然科学基金(62262028)
江西省教育厅科技课题(GJJ170268,GJJ200438,GJJ210438)资助项目.
关键词
协同过滤
推荐精度
可信度
可信预测值
鲁棒性
collaborative filtering
recommendation accuracy
credibility
credible predicted value
robustness
