信任关系辅助的隐反馈Web服务推荐研究
发布时间:2018-06-04 02:23
本文选题:信任知识 + 隐反馈 ; 参考:《武汉大学学报(理学版)》2017年02期
【摘要】:针对Web服务推荐现有技术缺乏显式打分数据缺点,提出使用隐反馈知识进行推荐的方法.该方法首先构造一个伪评分生成器,将用户隐反馈知识映射成为显式打分.基于矩阵因子分解模型,将信任知识引入服务推荐过程,建立一种融合社交信任信息的服务推荐模型,有效提高了服务推荐性能.实验表明,本文提出的基于隐反馈的服务推荐方法预测性能优于最近邻方法和SVD++方法;同SVD++方法的性能对比实验表明,引入信任知识能够进一步提高服务推荐的性能,具有较好的实际应用价值.
[Abstract]:In view of the lack of explicit data scoring in the current technology of Web service recommendation, a recommendation method using implicit feedback knowledge is proposed. Firstly, a pseudo-score generator is constructed to map the user's implicit feedback knowledge to explicit scoring. Based on the matrix factorization model, the trust knowledge is introduced into the service recommendation process, and a service recommendation model combining social trust information is established, which effectively improves the service recommendation performance. Experiments show that the performance of the proposed service recommendation method based on implicit feedback is superior to that of the nearest neighbor method and the SVD method, and the performance comparison with the SVD method shows that introducing trust knowledge can further improve the performance of service recommendation. It has good practical application value.
【作者单位】: 福建农林大学计算机与信息学院;武汉大学计算机学院;山东科技大学信息科学与工程学院;
【基金】:国家重点基础研究发展计划(973)(2014CB340600) 国家自然科学基金重点项目(6332019);国家自然科学基金资助项目(61173138,61272452) 福建省自然科学基金资助项目(2016J01285) 武汉大学软件工程国家重点实验室开放课题(SKLSE2014-10-07)
【分类号】:TP391.3;TP393.09
,
本文编号:1975406
本文链接:https://www.wllwen.com/guanlilunwen/ydhl/1975406.html