《计算机应用研究》|Application Research of Computers

基于多属性的动态采样协同过滤推荐算法

Collaborative filtering recommendation algorithm based on multi-attribute dynamic sampling

免费全文下载 (已被下载 次)  
获取PDF全文
作者 李维乾,张艺,郑振峰,王海,张紫云
机构 1.西安工程大学 计算机科学学院,西安 710048;2.陕西省服装设计智能化重点实验室,西安 710048;3.新型网络智能信息服务国家地方联合工程研究中心,西安 710048;4.陕西国防工业职业技术学院 电子工程学院,西安 710300;5.西北大学 信息科学与技术学院,西安 710127
统计 摘要被查看 次,已被下载
文章编号 1001-3695(2020)09-015-2640-05
DOI 10.19734/j.issn.1001-3695.2019.04.0124
摘要 针对目前协同过滤推荐算法推荐精度和用户数据在算法中匹配度都不高的问题,提出一种多属性的条件受限波尔兹曼机协同过滤推荐模型(MA-CRBM)。该模型基于实值状态的条件玻尔兹曼机,融合了用户职业和性别属性,充分利用数据集中潜在的评分与未评分信息。在训练过程中,采用动态迭代采样算法对原采样算法进行了改进,克服了训练后期数据采样误差波动太大导致精确度不高的问题。在MovieLens 数据集上的实验结果表明,MA-CRBM模型具有较好的推荐效果,可以有效提升推荐模型的精度和效率。
关键词 协同过滤推荐算法; 条件受限性玻尔兹曼机; 多属性条件推荐; 动态迭代采样算法
基金项目 国家自然科学基金资助项目(61572401,61672426,61701400)
西安工程大学博士科研启动基金资助项目(BS1330)
本文URL http://www.arocmag.com/article/01-2020-09-015.html
英文标题 Collaborative filtering recommendation algorithm based on multi-attribute dynamic sampling
作者英文名 Li Weiqian, Zhang Yi, Zheng Zhenfeng, Wang Hai, Zhang Ziyun
机构英文名 1.School of Computer Science,Xi'an Polytechnic University,Xi'an 710048,China;2.Shaanxi Key Laboratory of Clothing Intelligence,Xi'an 710048,China;3.State & Local Joint Engineering Research Center for Advanced Networking & Intelligent Information Services,Xi'an 710048,China;4.School of Electronic Engineering,Shaanxi Institute of Technology,Xi'an 710300,China;5.School of Information & Technology,Northwest University,Xi'an 710127,China
英文摘要 Aiming at the problem that the recommendation accuracy of collaborative filtering recommendation algorithm and the matching degree of user data in the algorithm are not high at present, this paper proposed a multi-attribute conditional restricted Boltzmann machine collaborative filtering recommendation model(MA-CRBM). The model was based on the conditional Boltzmann machine of real value state, which integrated the attributes of users' occupation and gender, and mode full use of the potential scoring and unscoring information in the data set. In the training process, it adopted the multi-step iterative dynamic sampling algorithm to improve the original sampling algorithm, which overcame the problem of low accuracy caused by too large fluctuation of data sampling error in the later training period. The experimental results on MovieLens data set show that the MA-CRBM model has a good recommendation effect and can effectively improve the accuracy and efficiency of the recommendation model.
英文关键词 collaborative filtering recommendation algorithm; conditionally constrained Boltzmann machine; multi-attribute condition recommendation; dynamic iterative sampling algorithm
参考文献 查看稿件参考文献
 
收稿日期 2019/4/17
修回日期 2019/6/19
页码 2640-2644,2683
中图分类号 TP393
文献标志码 A