计算机科学

计算机科学 ›› 2026, Vol. 53 ›› Issue (2): 152-160.doi: 10.11896/jsjkx.241200177

• 数据库&大数据&数据科学 • 上一篇    下一篇

基于用户静动态兴趣和去噪隐式负反馈的新闻推荐算法

魏金生1,2, 周苏1, 卢官明1,2, 丁佳伟1   

  1. 1 南京邮电大学通信与信息工程学院 南京 210003
    2 智能信息处理与通信技术省高校重点实验室(南京邮电大学) 南京 210003
  • 收稿日期:2024-12-24 修回日期:2025-05-10 发布日期:2026-02-10
  • 通讯作者: 卢官明(lugm@njupt.edu.cn)
  • 作者简介:(weijs@njupt.edu.cn)
  • 基金资助:
    国家自然科学基金(72074038);江苏省高等学校基础科学(自然科学)研究项目(24KJB520022);南京市留学人员科技创新项目(RK002NLX23004);南京邮电大学引进人才自然科学研究启动基金(NY223030);江苏省研究生科研与实践创新计划项目(KYCX22_0950)

News Recommendation Algorithm Based on User Static and Dynamic Interests and DenoisedImplicit Negative Feedback

WEI Jinsheng1,2, ZHOU Su1, LU Guanming1,2 , DING Jiawei1   

  1. 1 School of Communication and Information Engineering,Nanjing University of Posts and Telecommunications,Nanjing 210003,China
    2 Jiangsu Key Laboratory of Intelligent Information Processing and Communication Technology,Nanjing University of Posts and Telecommunications,Nanjing 210003,China
  • Received:2024-12-24 Revised:2025-05-10 Online:2026-02-10
  • About author:WEI Jinsheng,born in 1995,Ph.D,lecturer.His main research interests include multimedia information proce-ssing,machine learning,intelligent re-commendation and affective computing.
    LU Guanming,born in 1965,Ph.D,professor,Ph.D supervisor.His main research interests include intelligent information processing,intelligent recommendation and machine learning.
  • Supported by:
    National Natural Science Foundation of China(72074038),Natural Science Foundation of the Higher Education Institutions of Jiangsu Province(24KJB520022),Nanjing Science and Technology Innovation Foundation for Overseas Students (RK002NLX23004),Natural Science Research Start-up Foundation of Recruiting Talents of Nanjing University of Posts and Telecommunications(NY223030) and Postgraduate Research & Practice Innovation Program of Jiangsu Province(KYCX22_0950).

PDF (PC)

81

摘要： 在现有的新闻推荐系统中,通常将用户历史未点击新闻视为隐式负反馈,对隐式负反馈进行建模能够引导推荐模型过滤掉用户不感兴趣的新闻。未点击的新闻中可能会存在用户感兴趣的内容,即存在偏好噪声,导致对隐式负反馈建模的干扰;此外,由于用户兴趣的多样化和多变性,现有的新闻推荐系统往往存在“信息茧房”的问题。为了解决上述问题,提出了一种基于用户静动态兴趣和去噪隐式负反馈的新闻推荐算法。通过对用户的静态和动态兴趣进行融合建模,并对静动态兴趣隐式负反馈进行去噪,来构建动态更新的用户偏好模型。首先,构建基于正交映射的静态兴趣去噪模块,对静态兴趣中的隐式负反馈进行去噪;然后,将门控循环单元(Gated Recurrent Unit,GRU)与正交映射相融合,构建基于改进GRU的动态兴趣去噪模块,充分建模用户的兴趣变化并实现对动态兴趣中的隐式负反馈进行去噪;最后,通过引入对比学习技术,增强模型对隐式正反馈和负反馈的区分能力,以提升个性化新闻推荐性能。在MIND数据集上的实验表明,与基线方法相比,该模型在AUC,MRR,NDCG@5,NDCG@10这4种评价指标上分别提升了1.18%,1.84%,2.75%,1.67%,验证了该模型的有效性。

关键词: 新闻推荐, 信息茧房, 正交映射, 门控循环单元, 对比学习

Abstract: In the existing news recommendation system,the news that users have not clicked on in the past is usually regarded as implicit negative feedback,and modelling the implicit negative feedback can guide the recommendation model to filter out the news that users are not interested in.Because there may be content of interest to users in the news that is not clicked,that is,there is preference noise,which leads to interference with modelling implicit negative feedback.In addition,due to the diversification and variability of user interests,the existing news recommendation system often has the problem of an “information cocoon”.To solve the above problems,this paper proposes a news recommendation algorithm based on users’ static and dynamic interests and the denoised implicit negative feedback.By fusing and modeling the static and dynamic interests of users,and the denoised implicit negative feedback of static and dynamic interests,the dynamically updated user preference model is constructed.Firstly,a static interest denoising module based on orthogonal mapping is constructed to denoise the implicit negative feedback in static interest.Then,the GRU and orthogonal mapping are fused to construct a dynamic interest denoising module based on the improved GRU,which fully models the user’s interest change and realizes the denoising of the implicit negative feedback in the dynamic interest.Finally,by introducing contrastive learning technology,the model’s ability to distinguish between implicit positive and negative feedback is enhanced to improve the performance of personalized news recommendations.Experiments on the MIND dataset show that compared with the baseline method,the model improves by 1.18%,1.84%,2.75% and 1.67% on the four evaluation in-dexes of AUC,MRR,NDCG@5 and NDCG@10,respectively,which verifies the effectiveness of the proposed model.

Key words: News recommendation, Information cocoon, Orthogonal mapping, Gated recurrent unit, Contrastive learning

中图分类号: 

  • TP391
[1]HU Y F,QIU Z P,WU X.Denoising Neural Network for News Recommendation with Positive and Negative Implicit Feedback[C]//Findings of the Association for Computational Linguistics:NAACL 2022.Association for Computational Linguistics,2022:2320-2329.
[2]WU J L,CUI X H.News Recommendation Algorithm Based on Users’ Implicit Feedback Signals and Multi-Dimensional In-terests[J].Computer Engineering and Applications,2024,60(12):101-110.
[3]KOMPAN M,BIELIKOVÁ M.Content-based News Recom-mendation[C]//E-Commerce and Web Technologies:11th International Conference.Berlin:Springer,2010:61-72.
[4]DAS A S,DATAR M,GARG A,et al.Google News Personalization:Scalable Online Collaborative Filtering[C]//The 16th International Conference on World Wide Web.2007:271-280.
[5]LIU J,SONG J B,LI C,et al.A Hhybrid News Recommendation Algorithm based on K-means Clustering and Collaborative Filtering[J].Journal of Physics:Conference Series.IOP Publishing,2021,1881:032050.
[6]WANG H W,ZHANG F Z,XIE X,et al.DKN:Deep Know-ledge-aware Network for News Recommendation[C]//The 2018 World Wide Web Conference.2018:1835-1844.
[7]WU C H,WU F Z,AN M X,et al.NPA:Neural News Recommendation with Personalized Attention[C]//The 25th ACM SIGKDD International Conference on Knowledge Discovery & Data Mining.2019:2576-2584.
[8]WU C H,WU F Z,GE S Y,et al.Neural News Recommendation with Multi-Head Self-Attention[C]//The 2019 Conference on Empirical Methods in Natural Language Processing and the 9th International Joint Conference on Natural Language Processing(EMNLP-IJCNLP).2019:6389-6394.
[9]QI T,WU F Z,WU C H,et al.FUM:Fine-Grained and Fast User Modeling for News Recommendation[C]//Proceedings of the 45th International ACM SIGIR Conference on Research and Development in Information Retrieval.2022:1974-1978.
[10]TRAN D H,SHENG Q Z,ZHANG W E,et al. CupMar:ADeep Learning Model for Personalized News Recommendation Based on Contextual User-Profile and Multi-Aspect Article Representation[J].World Wide Web,2023,26(2):713-732.
[11]AN M X,WU F Z,WU C H,et al.Neural News Recommendation with Long-and Short-term User Representations[C]//The 57th Annual Meeting of the Association for Computational Linguistics.2019:336-345.
[12]QI T,WU F Z,WU C H,et al.News Recommendation with Candidate-Aware User Modeling[C]//Proceedings of the 45th International ACM SIGIR Conference on Research and Development in Information Retrieval.2022:1917-1921.
[13]WU C H,WU F Z,QI T,et al.Feedrec:News Feed Recommendation with Various User Feedbacks[C]//The ACM Web Conference 2022.2022:2088-2097.
[14]LIU J L,LIU T Q,YU C.Newsembed:Modeling News ThroughPre-trained Document Representations[C]//The 27th ACM SIGKDD Conference on Knowledge Discovery & Data Mining.2021:1076-1086.
[15]YANG Y H,HUANG C,XIA L H,et al.Knowledge Graph Contrastive Learning for Recommendation[C]//The 45th International ACM SIGIR Conference on Research and Development in Information Retrieval.2022:1434-1443.
[16]YU Y,WU F Z,WU C H,et al.Tiny-newsrec:Effective and Efficient PLM-based News Recommendation[C]//The 2022 Conference on Empirical Methods in Natural Language Processing.2022:5478-5489.
[17]WANG S C,GUO S,WANG L H,et al.Multi-Interest Extraction Joint with Contrastive Learning for News Recommendation[C]//Joint European Conference on Machine Learning and Knowledge Discovery in Databases.Cham:Springer,2022:606-621.
[18]SHIN K,KWAK H,KIM S Y,et al.Scaling Law for Recom-mendation Models:Towards General-purpose User Representations[C]//The AAAI Conference on Artificial Intelligence.2023:4596-4604.
[19]XIE R B,QIU Z J,ZHANG B,et al.Multi-Granularity Item-based Contrastive Recommendation[C]//International Confe-rence on Database Systems for Advanced Applications.Cham:Springer,2023:406-416.
[20]BIAN Z,ZHOU S J,FU H,et al.Denoising User-aware Memory Network for Recommendation[C]//The 15th ACM Conference on Recommender Systems.2021:400-410.
[21]OKURA S,TAGAMI Y,ONO S,et al.Embedding-based News Recommendation for Millions of Users[C]//The 23rd ACM SIGKDD International Conference on Knowledge Discovery and Data Mining.2017:1933-1942.
[22]WU F Z,QIAO Y,CHEN J H,et al.Mind:A Large-scale Dataset for News Recommendation[C]//The 58th Annual Meeting of the Association for Computational Linguistics.2020:3597-3606.
[23]PENNINGTON J,SOCHER R,MANNING C D.Glove:Global Vectors for Word Representation[C]//The 2014 Conference on Empirical Methods in Natural Language Processing(EMNLP).2014:1532-1543.
[24]RENDLE S.Factorization Machines with LibFM[J].ACMTransactions on Intelligent Systems and Technology,2012,3(3):1-22.
[25]GUO H F,TANG R M,YE Y M,et al.DeepFM:a Factorization-machine based Neural Network for CTR Prediction[C]//The 26th International Joint Conference on Artificial Intelligence(IJCAI’17).AAAI,2017:1725-1731.
Viewed
Full text


Abstract

Cited
  Shared   
  Discussed   
No Suggested Reading articles found!
渝ICP备16013121号-4
地址：重庆市两江新区洪湖西路18号    邮编：401121  
电话：023-63500828（编辑部） 023-67039612（会议/专题） 023-67039623（财务/发票）
E-mail：jsjkx12@163.com
本系统由北京玛格泰克科技发展有限公司设计开发