[1]申策,郝志峰,温雯,等.融合社交信息的跨域时序兴趣预测方法[J].郑州大学学报(工学版),2019,40(02):51-57.[doi:10.13705/j.issn.1671-6833.2019.02.024]
 Hao Zhifeng,Shen policy,Cai Ruichu,et al.A Cross-domain Temporal Interest Prediction Method by Integrating Social Information[J].Journal of Zhengzhou University (Engineering Science),2019,40(02):51-57.[doi:10.13705/j.issn.1671-6833.2019.02.024]
点击复制

融合社交信息的跨域时序兴趣预测方法()
分享到:

《郑州大学学报(工学版)》[ISSN:1671-6833/CN:41-1339/T]

卷:
40卷
期数:
2019年02期
页码:
51-57
栏目:
出版日期:
2019-03-19

文章信息/Info

Title:
A Cross-domain Temporal Interest Prediction Method by Integrating Social Information
作者:
申策郝志峰温雯蔡瑞初
广东工业大学计算机学院
Author(s):
Hao ZhifengShen policyCai RuichuWen Wen
School of Computer Science, Guangdong University of Technology
关键词:
兴趣预测跨域推荐社交信息时序行为排序学习
Keywords:
interest predictionCross domain recommendationsocial informationtiming behaviorsequence learning
DOI:
10.13705/j.issn.1671-6833.2019.02.024
文献标志码:
A
摘要:
引入用户的社交网络信息,是解决冷启动问题提升推荐效果的有效方式.针对现有引入社交关系的静态方法忽略用户兴趣的变化导致预测结果存在滞后性,提出一种跨域时序兴趣预测CDTIP方法,引入用户社交行为信息.首先提出跨域个性化排序模型,实现社交特征和购物特征的跨域融合.其次按时间段划分用户历史行为,提出一种时序特征建模方法.在真实数据集上的验证结果表明,所提出的方法能更有效地预测用户兴趣.
Abstract:
Integrating user’s social information is an appropriate way to solve the user-cold start problem. Though various prediction models focus on integrating social relation information, few have noticed the dynamic change of the user’s interest. Thus, in this paper, we propose a cross-domain temporal interest prediction approach by integrating social activity information. First, we construct a cross-domain personized ranking model which can map the feature from social space into the purchase space. Further, we propose a feature modeling method based on data grouped by time period. Experiments on the dataset verify that the proposed method can predict user’s interest more effectively.
更新日期/Last Update: 2019-03-24