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

科研社交网络中基于异质网络分析的列表级排序学习推荐方法研究

Study of listwise learning-to-rank recommendation method based on heterogeneous network analysis in scientific social network

免费全文下载 (已被下载 次)  
获取PDF全文
作者 岳峰,王含茹,张馨悦,王刚
机构 合肥工业大学 a.计算机与信息学院;b.管理学院,合肥 230009
统计 摘要被查看 次,已被下载
文章编号 1001-3695(2020)12-005-3552-05
DOI 10.19734/j.issn.1001-3695.2019.09.0538
摘要 针对现有学术论文推荐方法不能充分利用科研社交网络中实体间的异质关系,且大多聚焦于预测评分的准确性,忽略用户偏好顺序的问题,提出一种基于异质网络分析的列表级排序学习推荐方法。首先采用异质网络分析充分探究科研社交网络中实体之间的关系,在此基础上将异质网络分析获取的信息融入列表级排序学习框架中,对学术论文的推荐排序列表进行优化,最终得到为科研人员推荐的学术论文列表。在科研社交网络科研之友数据集上的实验结果表明所提方法较其他传统推荐方法取得了更好的结果,验证了该方法的有效性。
关键词 科研社交网络; 论文推荐; 异质网络; 列表级排序学习
基金项目 国家自然科学基金资助项目(71471054,91646111)
教育部人文社科基金资助项目(18YJC870025)
安徽省自然科学基金资助项目(1608085MG150)
本文URL http://www.arocmag.com/article/01-2020-12-005.html
英文标题 Study of listwise learning-to-rank recommendation method based on heterogeneous network analysis in scientific social network
作者英文名 Yue Feng, Wang Hanru, Zhang Xinyue, Wang Gang
机构英文名 a.School of Computer Science & Information Engineering,b.School of Management,Hefei University of Technology,Hefei 230009,China
英文摘要 In view of the fact that the existing recommendation methods for academic papers cannot make full use of the hete-rogeneous relations between entities in scientific social network, and most of them focus on the accuracy of predicted ratings, ignoring users' preference order, this paper proposed a heterogeneous network analysis based listwise learning-to-rank method. Firstly, it employed heterogeneous network analysis to fully explore the complex relations between entities in the scientific social network. On this basis, it integrated the information obtained from heterogeneous network analysis into the listwise method to optimize the ranking of the papers, which finally got the recommendation list of papers to researchers. This paper conducted experiments on the dataset of ScholarMate(one of the prevalent scientific social networks). And the experimental results show that the proposed method performs better than traditional recommendation methods, which illustrates the effectiveness of the proposed method.
英文关键词 scientific social network; paper recommendation; heterogeneous network; listwise learning-to-rank
参考文献 查看稿件参考文献
 
收稿日期 2019/9/19
修回日期 2019/11/9
页码 3552-3556,3564
中图分类号 TP391.3
文献标志码 A