基于时间特征的图卷积新闻推荐模型

doi:10.3778/j.issn.1673-9418.2501019

摘要/Abstract

摘要： 针对现有新闻推荐系统在动态特征提取中未能有效挖掘用户交互数据与阅读习惯信息的问题，提出一种基于时间特征的图卷积网络模型TimelyGCN。该模型充分考虑用户停留时长、阅读行为的时序特征以及新闻的生命周期属性，以刻画用户兴趣的演化和新闻内容的新鲜度变化。在新闻特征提取阶段，引入阅读时长数据，并结合时态建模与语义分析，深度挖掘用户与新闻内容的动态关联。同时，构建用户-新闻交互图，并在预训练阶段将行为时间序列特征与新闻生命周期属性作为复合权重融入图结构中，以显式建模用户兴趣的时间依赖性，从而增强模型对新闻消费模式的适应能力。在推荐预测阶段，进一步结合新闻的剩余生命周期信息，采用动态权重调控机制优化候选新闻的曝光优先级，以平衡个性化推荐与新闻内容时效性之间的关系，从而提升推荐效果的精准度和实时性。实验结果显示，提出的模型在Adressa数据集上有效提升了推荐性能，相较于现有基线网络方法，在AUC、MRR、nDCG@5和nDCG@10指标上均取得了不同程度的提升。

关键词: 图卷积网络, 交互图, 时间感知, 动态偏好

Abstract: To address the issue that existing news recommendation systems fail to effectively extract user interaction data and reading habit information in dynamic feature extraction, this paper proposes a time-aware graph convolutional network model, TimelyGCN. This model comprehensively considers user dwell time, the temporal characteristics of reading behavior, and the lifecycle attributes of news articles to capture the evolution of user interests and the freshness changes of news content. During the news feature extraction phase, reading duration data are incorporated, and temporal modeling along with semantic analysis is used to deeply explore the dynamic relationship between users and news content. Furthermore, this paper constructs a user-news interaction graph and integrates behavior time-series features and news lifecycle attributes as composite weights into the graph structure during the pretraining phase, explicitly modeling the temporal dependencies of user interests, thereby enhancing the model’s adaptability to news consumption patterns. In the recommendation prediction phase, residual lifecycle information of news articles is further leveraged, and a dynamic weight adjustment mechanism is employed to optimize the exposure priority of candidate news, balancing personalized recommendations with the timeliness of news content, thus improving the accuracy and timeliness of recommendations. Experimental results demonstrate that the proposed model significantly enhances recommendation performance on the Adressa dataset, achieving improvements across various metrics, including AUC, MRR, nDCG@5, and nDCG@10, compared with existing baseline methods.

Key words: graph convolutional network, interaction graph, temporal awareness, dynamic preferences

杨智勇, 陈佳慧, 许沁欣. 基于时间特征的图卷积新闻推荐模型[J]. 计算机科学与探索, 2025, 19(11): 3059-3071.

YANG Zhiyong, CHEN Jiahui, XU Qinxin. Graph Convolutional News Recommendation Model Based on Temporal Features[J]. Journal of Frontiers of Computer Science and Technology, 2025, 19(11): 3059-3071.

参考文献

[1] 孟祥福, 霍红锦, 张霄雁, 等. 个性化新闻推荐方法研究综述[J]. 计算机科学与探索, 2023, 17(12): 2840-2860.
MENG X F, HUO H J, ZHANG X Y, et al. Survey of research on personalized news recommendation approaches[J]. Journal of Frontiers of Computer Science and Technology, 2023, 17(12): 2840-2860.
[2] WU C H, WU F Z, QI T, et al. FeedRec: news feed recommendation with various user feedbacks[C]//Proceedings of the ACM Web Conference 2022. New York: ACM, 2022: 2088-2097.
[3] WU C H, WU F Z, GE S Y, et al. Neural news recommendation with multi-head self-attention[C]//Proceedings of the 2019 Conference on Empirical Methods in Natural Language Processing and the 9th International Joint Conference on Natural Language Processing. Stroudsburg: ACL, 2019: 6388-6393.
[4] NASERI M, NASERI M, ZAMANI H, et al. Analyzing and predicting news popularity in an instant messaging service[C]//Proceedings of the 42nd International ACM SIGIR Conference on Research and Development in Information Retrieval. New York: ACM, 2019: 1053-1056.
[5] LIAO D, XU J, LI G, et al. Popularity prediction on online articles with deep fusion of temporal process and content features[C]//Proceedings of the 33rd AAAI Conference on Artificial Intelligence and the 31st Innovative Applications of Artificial Intelligence Conference and the 9th AAAI Symposium on Educational Advances in Artificial Intelligence. Palo Alto: AAAI, 2019: 200-207.
[6] WU C H, WU F Z, AN M X, et al. Neural news recommendation with topic-aware news representation[C]//Proceedings of the 57th Annual Meeting of the Association for Computational Linguistics. Stroudsburg: ACL, 2019: 1154-1159.
[7] HU L M, LI C, SHI C, et al. Graph neural news recommendation with long-term and short-term interest modeling[J]. Information Processing & Management, 2020, 57(2): 102142.
[8] MA M Y, NA S, WANG H Y, et al. The graph-based behavior-aware recommendation for interactive news[J]. Applied Intelligence, 2022, 52(2): 1913-1929.
[9] XIE R, LING C, WANG Y, et al. Deep feedback network for recommendation[C]//Proceedings of the 29th International Joint Conference on Artificial Intelligence. Palo Alto: AAAI, 2021: 2519-2525.
[10] YI X, HONG L J, ZHONG E H, et al. Beyond clicks: dwell time for personalization[C]//Proceedings of the 8th ACM Conference on Recommender Systems. New York: ACM, 2014: 113-120.
[11] XIE R, MA L, ZHANG S, et al. Reweighting clicks with dwell time in recommendation[C]//Proceedings of the ACM Web Conference 2023. New York: ACM, 2023: 341-345.
[12] JI Z Y, WU M D, YANG H, et al. Temporal sensitive heterogeneous graph neural network for news recommendation[J]. Future Generation Computer Systems, 2021, 125: 324-333.
[13] LI J Y, ZHANG Y, LIN X, et al. TAML: time-aware meta learning for cold-start problem in news recommendation[C]//Proceedings of the 46th International ACM SIGIR Conference on Research and Development in Information Retrieval. New York: ACM, 2023: 2415-2419.
[14] JIANG H, LI C Z, AN M X. Time matters: enhancing pre-trained news recommendation models with robust user dwell time injection[EB/OL]. [2024-11-09]. https://arxiv.org/abs/2405.12486.
[15] WU C, WU F, QI T, et al. User modeling with click preference and reading satisfaction for news recommendation[C]//Proceedings of the 29th International Joint Conference on Artificial Intelligence. Palo Alto: AAAI, 2020: 3023-3029.
[16] WU C H, WU F Z, QI T, et al. Quality-aware news recommendation[EB/OL]. [2024-11-09]. https://arxiv.org/abs/2202. 13605.
[17] WU C H, WU F Z, HUANG Y F, et al. Neural news recommendation with negative feedback[J]. CCF Transactions on Pervasive Computing and Interaction, 2020, 2(3): 178-188.
[18] 曹璐, 丁苍峰, 马乐荣, 等. 面向图神经网络的节点重要性排序研究进展[J]. 计算机科学与探索, 2025, 19(4): 877-900.
CAO L, DING C F, MA L R, et al. Advances in node importance ranking based on graph neural networks[J]. Journal of Frontiers of Computer Science and Technology, 2025, 19(4): 877-900.
[19] YING R, HE R N, CHEN K F, et al. Graph convolutional neural networks for web-scale recommender systems[C]//Proceedings of the 24th ACM SIGKDD International Conference on Knowledge Discovery and Data Mining. New York: ACM, 2018: 974-983.
[20] DARVISHY A, IBRAHIM H, SIDI F, et al. HYPNER: a hybrid approach for personalized news recommendation[J]. IEEE Access, 2020, 8: 46877-46894.
[21] WANG X, HE X, WANG M, et al. Neural graph collaborative filtering[C]//Proceedings of the 42nd International ACM SIGIR Conference on Research and Development in Information Retrieval. New York: ACM, 2019: 165-174.
[22] QI T, WU F Z, WU C H, et al. HieRec: hierarchical user interest modeling for personalized news recommendation[EB/OL]. [2024-11-10]. https://arxiv.org/abs/2106.04408.
[23] CHEN L, WU L, HONG R C, et al. Revisiting graph based collaborative filtering: a linear residual graph convolutional network approach[J]. Proceedings of the AAAI Conference on Artificial Intelligence, 2020, 34(1): 27-34.
[24] HE X N, DENG K, WANG X, et al. LightGCN: simplifying and powering graph convolution network for recommendation[C]//Proceedings of the 43rd International ACM SIGIR Conference on Research and Development in Information Retrieval. New York: ACM, 2020: 639-648.
[25] LIAO J, ZHOU W, LUO F J, et al. SocialLGN: light graph convolution network for social recommendation[J]. Information Sciences, 2022, 589: 595-607.
[26] HAMILTON W, YING Z, LESKOVEC J. Inductive representation learning on large graphs[C]//Proceedings of the 31st International Conference on Neural Information Processing Systems, 2017: 1024-1034.
[27] CHIANG W L, LIU X Q, SI S, et al. Cluster-GCN: an efficient algorithm for training deep and large graph convolutional networks[C]//Proceedings of the 25th ACM SIGKDD International Conference on Knowledge Discovery and Data Mining. New York: ACM, 2019: 257-266.
[28] PAREJA A, DOMENICONI G, CHEN J, et al. EvolveGCN: evolving graph convolutional networks for dynamic graphs[J]. Proceedings of the AAAI Conference on Artificial Intelligence, 2020, 34(4): 5363-5370.
[29] RAZA S, DING C. A survey on news recommender system-dealing with timeliness, dynamic user interest and content quality, and effects of recommendation on news readers[EB/OL]. [2024-11-10]. https://arxiv.org/abs/2009.04964.
[30] WU C H, WU F Z, HUANG Y F, et al. Personalized news recommendation: methods and challenges[J]. ACM Transactions on Information Systems, 2023, 41(1): 1-50.
[31] WU F Z, QIAO Y, CHEN J H, et al. MIND: a large-scale dataset for news recommendation[C]//Proceedings of the 58th Annual Meeting of the Association for Computational Linguistics. Stroudsburg: ACL, 2020: 3597-3606.
[32] GULLA J A, ZHANG L, LIU P, et al. The Adressa dataset for news recommendation[C]//Proceedings of the 2017 International Conference on Web Intelligence. New York: ACM, 2017: 1042-1048.
[33] HUANG Z Y, JIN B B, ZHAO H K, et al. Personal or general? A hybrid strategy with multi-factors for news recommendation[J]. ACM Transactions on Information Systems, 2023, 41(2): 1-29.
[34] BAE H K, AHN J, LEE D, et al. LANCER: a lifetime-aware news recommender system[J]. Proceedings of the AAAI Conference on Artificial Intelligence, 2023, 37(4): 4141-4148.
[35] GE S, WU C, WU F, et al. Graph enhanced representation learning for news recommendation[C]//Proceedings of the Web Conference 2020. New York: ACM, 2020: 2863-2869.
[36] MAO Z M, LI J, WANG H R, et al. DIGAT: modeling news recommendation with dual-graph interaction[EB/OL]. [2024-11-10]. https://arxiv.org/abs/2210.05196.
[37] JONNALAGEDDA N, GAUCH S, LABILLE K, et al. Incorporating popularity in a personalized news recommender system[J]. PeerJ Computer Science, 2016, 2: e63.
[38] DEVLIN J, CHANG M W, LEE K, et al. BERT: pre-training of deep bidirectional transformers for language understanding[EB/OL]. [2024-11-12]. https://arxiv.org/abs/1810.04805.
[39] WANG J K, JIANG Y T, LI H C, et al. Improving news recommendation with channel-wise dynamic representations and contrastive user modeling[C]//Proceedings of the 16th ACM International Conference on Web Search and Data Mining. New York: ACM, 2023: 562-570.