计算机工程 ›› 2019, Vol. 45 ›› Issue (5): 175-181.doi: 10.19678/j.issn.1000-3428.0050495

• 人工智能及识别技术 • 上一篇    下一篇

基于多重图排序的用户冷启动推荐方法

毛明松,张富国   

  1. 江西财经大学 信息管理学院,南昌 330013
  • 收稿日期:2018-02-15 出版日期:2019-05-15 发布日期:2019-05-15
  • 作者简介:毛明松(1987—),男,讲师、博士,主研方向为个性化推荐、数据挖掘;张富国,副教授、博士。
  • 基金项目:

    国家自然科学基金(71361012,71764006);江西省自然科学基金(20161BAB201029);江西省教育厅科技项目(GJJ170344,GJJ170319)。

Recommendation Method for User Cold-start Based on Multi-graph Rank

MAO Mingsong,ZHANG Fuguo   

  1. School of Information Management,Jiangxi University of Finance and Economics,Nanchang 330013,China
  • Received:2018-02-15 Online:2019-05-15 Published:2019-05-15

摘要:

为使用户-物品评分、社会网络和社会化标签等异构信息融合到协同过滤推荐方法的最近邻寻找过程中,弥补冷启动用户单一维度信息的不足,提出一种多重图排序的冷启动推荐方法。通过分析用户之间可能存在的信任度构建关系网络,利用多重图排序模型得到目标用户的最近邻集合,进而产生目标用户的推荐列表。实验结果表明,与基于用户的协同过滤推荐方法相比,该方法能有效地提高冷启动用户的个性化推荐准确性和推荐覆盖率。

关键词: 推荐方法, 异构信息, 冷启动, 多重图模型, 多元关系网络

Abstract:

In order to integrate the heterogeneous information such as user-item scoring,social network and social labeling into the nearest neighbor search process of collaborative filtering recommendation method,and to make up for the lack of information on the single dimension of cold-start users a cold-start recommendation method for multi-graph rank is proposed.A relational network is built by analyzing the trustworthiness that may exist between users,and the multi-graph sorting model is used to obtain the nearest neighbor set of the target user,thereby generating a recommendation list of the target user.Experimental results show that compared with the user-based collaborative filtering recommendation method,this method can effectively improve the personalized recommendation accuracy and recommended coverage rate of cold-start users.

Key words: recommendation method, heterogeneous information, cold start, multi-graph model, multi-relational network

中图分类号: