面向序列推荐的扩散增强多视角意图对比学习方法

doi:10.3778/j.issn.1002-8331.2410-0170

摘要/Abstract

摘要： 针对意图驱动的序列推荐中出现的数据稀疏性和意图信息提取不足问题，提出了一种基于扩散增强的多视角意图对比学习方法（DMVRec）。为学习原始项目间的购买意图相关性，在扩散模型反向过程中，通过融合原始序列意图信息和无分类器指导策略来预测高斯噪声。利用原始项目间的购买意图相关性信息指导反向过程的去噪，以得到增强序列的意图预测值，并利用余弦相似度计算每个位置下的增强项目。通过切分用户交互序列和聚类算法来构建意图监督信号，并分别在宏观和微观视角下利用意图对比学习来捕捉隐藏的用户意图信息。通过在四个公开数据集下实验，充分体现了DMVRec方法的优越性。其中，对比最先进基线，在HR@5指标上，提升7.03%，在NDCG@5指标上，提升4.53%。

关键词: 序列推荐, 扩散增强, 对比学习, 多视角意图

Abstract: To address data sparsity and inadequate intent extraction in intent-driven sequential recommendation, a diffusion-augmented multi-view intent contrastive learning method (DMVRec） is proposed. First, in the reverse diffusion process, Gaussian noise is predicted by integrating intent information from the original sequence with a classifier-free guidance strategy to learn purchasing intent correlations between items. Second, the denoising process leverages these correlations to generate intent predictions for the augmented sequence, with cosine similarity used to compute augmented items at each position. Finally, intent supervision signals are constructed by segmenting user interaction sequences and applying clustering algorithms, capturing hidden user intents through contrastive learning from both macro and micro perspectives. Experiments on four public datasets demonstrate the superiority of DMVRec, showing a 7.03% improvement in HR@5 and a 4.53% improvement in NDCG@5 over state-of-the-art baselines.

Key words: sequential recommendation, diffusion augmentation, contrastive learning, multi-view intent

王澳飞, 孙福振, 孙秀娟, 张文轩, 王绍卿. 面向序列推荐的扩散增强多视角意图对比学习方法[J]. 计算机工程与应用, 2025, 61(13): 338-348.

WANG Aofei, SUN Fuzhen, SUN Xiujuan, ZHANG Wenxuan, WANG Shaoqing. Diffusion-Augmented Multi-View Intent Contrastive Learning Method for Sequential Recommendation[J]. Computer Engineering and Applications, 2025, 61(13): 338-348.

参考文献

[1] 张惠鹃, 黄钦阳, 胡诗彦, 等. 完全图高阶关系驱动的链接预测[J]. 计算机研究与发展, 2024, 61(7）: 1825-1835.
ZHANG H J, HUANG Q Y, HU S Y, et al. Link prediction driven by high-order relations in complete graph[J]. Journal of Computer Research and Development, 2024, 61(7）: 1825-1835.
[2] 魏楚元, 袁保杰, 王昌栋. 多层级用户兴趣与多意图融合的下一篮推荐算法[J]. 计算机科学与探索, 2025, 19(3）: 749-763.
WEI C Y, YUAN B J, WANG C D. Multi-level user interest and multi-intent fusion for next basket recommendation[J]. Journal of Frontiers of Computer Science and Technology, 2025, 19(3）: 749-763.
[3] BOKA T F, NIU Z D, NEUPANE R B. A survey of sequential recommendation systems: techniques, evaluation, and future directions[J]. Information Systems, 2024, 125: 102427.
[4] RENDLE S, FREUDENTHALER C, SCHMIDT-THIEME L. Factorizing personalized Markov chains for next-basket recommendation[C]//Proceedings of the 19th International Conference on World Wide Web. New York: ACM, 2010: 811-820.
[5] JANNACH D, LUDEWIG M. When recurrent neural networks meet the neighborhood for session-based recommendation[C]//Proceedings of the 11th ACM Conference on Recommender Systems. New York: ACM, 2017: 306-310.
[6] KANG W C, MCAULEY J. Self-attentive sequential recommendation[C]//Proceedings of the 2018 IEEE International Conference on Data Mining. Piscataway: IEEE, 2018: 197-206.
[7] SUN F, LIU J, WU J, et al. BERT4Rec: sequential recommendation with bidirectional encoder representations from transformer[C]//Proceedings of the 28th ACM International Conference on Information and Knowledge Management. New York: ACM, 2019: 1441-1450.
[8] 田志轩, 刘骊, 付晓东, 等. 融合偏好学习和意图建模的个性化服装序列推荐[J/OL]. 计算机工程: 1-10 (2024-09-04） [2024-10-08]. https://doi.org/10.19678/j.issn.1000-3428.0069819.
TIAN Z X, LIU L, FU X D, et al. Personalized seq-uential clothing recommendation combining preference learning and intention modeling[J/OL]. Computer Engineering: 1-10 (2024-09-04） [2024-10-08]. https://doi.org/10.19678/j.issn.10003428. 0069819.
[9] CHEN Y J, LIU Z W, LI J, et al. Intent contrastive learning for sequential recommendation[C]//Proceedings of the ACM Web Conference 2022. New York: ACM, 2022: 2172-2182.
[10] TAN Q Y, ZHANG J W, YAO J C, et al. Sparse-interest network for sequential recommendation[C]//Proceedings of the 14th ACM International Conference on Web Search and Data Mining. New York: ACM, 2021: 598-606.
[11] LI X W, SUN A T, ZHAO M K, et al. Multi-intention oriented contrastive learning for sequential recommendation[C]//Proceedings of the Sixteenth ACM International Conference on Web Search and Data Mining. New York: ACM, 2023: 411-419.
[12] QIN X Y, YUAN H H, ZHAO P P, et al. Intent contrastive learning with cross subsequences for sequential recommendation[C]//Proceedings of the 17th ACM International Conference on Web Search and Data Mining. New York: ACM, 2024: 548-556.
[13] 刘会, 张璇, 杨兵, 等. 用于社交推荐的增强影响扩散模型[J]. 计算机学报, 2023, 46(3）: 626-642.
LIU H, ZHANG X, YANG B, et al. An enhanced influence diffusion model for social recommendation[J]. Chinese Journal of Computers, 2023, 46(3）: 626-642.
[14] LI J C, ZHAO T, LI J, et al. Coarse-to-fine sparse sequential recommendation[C]//Proceedings of the 45th International ACM SIGIR Conference on Research and Development in Information Retrieval. New York: ACM, 2022: 2082-2086.
[15] MA J X, ZHOU C, YANG H X, et al. Disentangled self-supervision in sequential recommenders[C]//Proceedings of the 26th ACM SIGKDD International Conference on Knowledge Discovery & Data Mining. New York: ACM, 2020: 483-491.
[16] XIE X, SUN F, LIU Z Y, et al. Contrastive learning for sequential recommendation[C]//Proceedings of the 2022 IEEE 38th International Conference on Data Engineering. Piscataway: IEEE, 2022: 1259-1273.
[17] LIU Z W, CHEN Y J, LI J, et al. Contrastive self-supervised sequential recommendation with robust augmentation[J]. arXiv:2108.06479, 2021.
[18] QIU R H, HUANG Z, YIN H Z, et al. Contrastive learning for representation degeneration problem in sequential recommendation[C]//Proceedings of the Fifteenth ACM International Conference on Web Search and Data Mining. New York: ACM, 2022: 813-823.
[19] CHEN J Y, ZOU G X, ZHOU P, et al. Sparse enhanced network: an adversarial generation method for robust augmentation in sequential recommendation[J]. Proceedings of the AAAI Conference on Artificial Intelligence, 2024, 38(8）: 8283-8291.
[20] LIU Z W, FAN Z W, WANG Y, et al. Augmenting sequential recommendation with pseudo-prior items via reversely pre-training transformer[C]//Proceedings of the 44th International ACM SIGIR Conference on Research and Development in Information Retrieval. New York: ACM, 2021: 1608-1612.
[21] LIU Q, YAN F, ZHAO X, et al. Diffusion augmentation for sequential recommendation[C]//Proceedings of the 32nd ACM International Conference on Information and Knowledge Management. New York: ACM, 2023: 1576-1586.
[22] 张文龙, 孙福振, 吴相帅, 等. 基于反向延长增强的对抗生成网络推荐算法[J]. 计算机应用研究, 2024, 41(7）: 2033-2038.
ZHANG W L, SUN F Z, WU X S, et al. Generative adversarial network recommendation based on reverse extension enhancement[J]. Application Research of Computers, 2024, 41(7）: 2033-2038.
[23] KINGMA D P, WELLING M. Auto-encoding variational Bayes[J]. arXiv:1312.6114, 2013.
[24] HO J, SALIMANS T. Classifier-free diffusion guidance[J]. arXiv:2207.12598, 2022.
[25] HO J, JAIN A, ABBEEL P. Denoising diffusion probabilistic models[C]//Advances in Neural Information Processing Systems, 2020: 6840-6851.
[26] 温民伟, 梅红岩, 袁凤源, 等. 多任务推荐算法研究综述[J]. 计算机科学与探索, 2024, 18(2）: 363-377.
WEN M W, MEI H Y, YUAN F Y, et al. Survey of multi-task recommendation algorithms[J]. Journal of Frontiers of Computer Science and Technology, 2024, 18(2）: 363-377.
[27] 王永, 罗陈红, 邓江洲, 等. 基于目标扰动的本地化差分隐私矩阵分解推荐算法[J]. 计算机学报, 2025, 48(2）: 451-462.
WANG Y, LUO C H, DENG J Z, et al. Matrix factorization recommendation algorithm based on local differential privacy with objective perturbation[J]. Chinese Journal of Computers, 2025, 48(2）: 451-462.
[28] TANG J X, WANG K. Personalized top-N sequential recommendation via convolutional sequence embedding[C]//Proceedings of the Eleventh ACM International Conference on Web Search and Data Mining. New York: ACM, 2018: 565-573.
[29] YUE Z R, WANG Y Q, HE Z K, et al. Linear recurrent units for sequential recommendation[C]//Proceedings of the 17th ACM International Conference on Web Search and Data Mining. New York: ACM, 2024: 930-938.