基于边缘计算的融合多因素的个性化推荐算法
Personalized recommendation algorithm fusing multiple factors in edge computing
投稿时间:2021-12-02  修订日期:2021-12-02
DOI:
中文关键词: 边缘计算  个性化推荐  多因素  数据稀疏  冷启动
英文关键词: edge computing  personalized recommendation  multiple factor  data sparsity  cold start
基金项目:国家重点基础研究发展计划(973计划),国家自然科学基金项目(面上项目,重点项目,重大项目),中央高校基本科研业务费专项资金项目
作者单位邮编
金焕章 中南民族大学 
朱容波 中南民族大学 430070
刘浩 中南民族大学 
陈慧敏 中南民族大学 
摘要点击次数: 95
全文下载次数: 0
中文摘要:
      针对传统推荐算法使用单一上下文信息不能有效的解决信息过载、数据稀疏、冷启动问题,本文提出了一种F-SVD算法(基于边缘计算的融合多因素的个性化推荐算法),和新的用户相似度计算方法F-PEARSON(改进后的PEARSON相关系数).在边缘服务器处理个性化用户数据以分散云服务器的压力,目前大多采用的集中式处理方式无法在爆炸性式增长的数据下提供准确及时的推荐,在云服务器融合多因素挖掘用户之间的潜在关系,从而构建预测算法F-SVD.实验结果表明,在公开数据集MovieLens上,与传统算法相比,本文算法在RMSE和MAE上的误差更小,精确度提升了2.2%.
英文摘要:
      For the traditional recommendation algorithm using single contextual information cannot effectively solve the problems of information overload, data sparsity, and cold start, this paper proposes an F-SVD algorithm (a personalized recommendation algorithm based on edge computing fused with multiple factors), and a new user similarity calculation method F-PEARSON (improved PEARSON correlation coefficient), using a machine learning-based BERT model for user history data training. Processing personalized user data at the edge server disperses the pressure of the cloud server, and the centralized processing method mostly used now cannot provide accurate and timely recommendations undesr the explosive growth of data, and the potential relationship between users is mined by fusing multiple factors at the cloud server, thus building the prediction algorithm F-SVD. experimental results show that on the public dataset MovieLens, compared with the traditional algorithm The algorithm in this paper has smaller errors on RMSE and MAE, and the accuracy is improved by 2.2%.
View Fulltext   查看/发表评论  下载PDF阅读器
关闭