基于双向解耦知识迁移的跨域推荐算法

Dual Disentangled Knowledge Transfer for Cross-Domain Recommendation

摘要

跨域推荐已经被证明在解决传统推荐系统中的长期存在的数据稀疏和冷启动问题方面非常有效. 然而，现有的方法迁移知识的方式是粗粒度的，忽略了交互背后的多种潜在因素. 为此，本研究设计名为 D³-CDR 的跨域推荐模型，用于在域之间双向迁移解耦知识. 具体而言，本研究设计双向解耦知识迁移层，该层由多个解耦知识迁移(DKT)块组成. 在每个 DKT 块中，通过将交互表示投射到多个潜在语义空间中生成解耦表示. 本研究提出的注意力知识迁移机制可以提取有用的知识，生成跨域的知识表示，自适应知识融合机制进一步自适应地融合域内的知识与跨域的知识. 在两个真实世界数据集上的大量实验表明，本研究提出的 D³-CDR 模型具有更优越的推荐性能.

Abstract

Cross-domain recommendation (CDR) has been proved to be effective to address the long-standing data sparsity and cold-start problems in traditional recommender systems. However, existing methods transfer knowledge at a coarse-grained level, ignoring multiple latent factors behind interactions. To this end, we propose a cross-domain recommendation model named D³-CDR, which enables dual knowledge transfer between domains. Specifically, we design a dual disentangled knowledge transfer layer that consists of several disentangled knowledge transfer (DKT) blocks. In each DKT block, we generate disentangled representations by projecting interaction representations into multiple latent semantic spaces. The attentive knowledge transfer mechanism distills useful knowledge generating cross-domain knowledge representations. The adaptive knowledge fusion mechanism further fuses within-domain knowledge with cross-domain knowledge adaptively. Extensive experiments on two real-world datasets demonstrate the effectiveness of our proposed D³CDR model.

关键词

推荐系统 / 跨域推荐 / 解耦表示学习 / 注意力知识迁移 / 自适应知识融合

Key words

recommendation system / cross-domain recommendation / disentangled representation learning / attentive knowledge transfer / adaptive knowledge fusion

参考文献

[1] He M, Zhang J, Yang P, Yao K. Robust transfer learning for cross—domain collaborative filtering using multiple rating patterns approximation[C] //Proceedings of the Eleventh ACM International Conference on Web Search and Data Mining. Marina Del Rey, CA, USA: ACM, 2018: 225-233.

[2] Li B, Yang Q, Xue X. Can movies and books collaborate? Cross—domain collaborative filtering for sparsity reduction[C] //Proceedings of the 21st International Joint Conference on Artificial Intelligence. Pasadena, CA, USA: IJCAI, 2009: 2052-2057.

[3] Li L, Do Q, Liu W. Cross—domain recommendation via coupled factorization machines[C] //Proceedings of the Thirty—Third AAAI Conference on Artificial Intelligence. Honolulu, Hawaii, USA: AAAI Press, 2019: 9965-9966.

[4] Yang Z, Peng Z, Wang Z, et al. FedGCDR: federated graph learning for cross—domain recommendation[ C] //Advances in Neural Information Processing Systems 38. Vancouver, BC, Canada: Curran Associates, Inc., 2024.

[5] Zhu F, Chen C, Wang Y, et al. DTCDR: a framework for dual—target cross—domain recommendation[C] //Proceedings of the 28th ACM International Conference on Information and Knowledge Management. Beijing, China: ACM, 2019: 1533-1542.

[6] Hu G, Zhang Y, Yang Q. CoNet: collaborative cross networks for cross—domain recommendation[C] //Proceedings of the 27th ACM International Conference on Information and Knowledge Management. Torino, Italy: ACM, 2018: 667-676.

[7] Liu J, Zhao P, Zhuang F, et al. Exploiting aesthetic preference in deep cross networks for cross—domain recommendation[C] //Proceedings of The Web Conference 2020. Taipei, China: ACM, 2020: 2768-2774.

[8] Liu W, Chen C, Liao X, et al. User distribution mapping modelling with collaborative filtering for cross domain recommendation[C] //Proceedings of the ACM Web Conference 2024. Singapore: ACM, 2024: 334-343.

[9] Liu M, Li J, Li G, et al. Cross domain recommendation via bi—directional transfer graph collaborative filtering networks[C] //Proceedings of the 29th ACM International Conference on Information and Knowledge Management. Virtual Event, Ireland: ACM, 2020: 885-894.

[10] Du J, Ye Z, Guo B, et al. Identifiability of cross—domain recommendation via causal subspace disentanglement [ C] //Proceedings of the 47th International ACM SIGIR Conference on Research and Development in Information Retrieval. Washington, DC, USA: ACM, 2024: 2091-2101.

[11] Chen H, Xin X, Wang D, et al. Decomposed collaborative filtering: modeling explicit and implicit factors for recommender systems[C] //Proceedings of the Fourteenth ACM International Conference on Web Search and Data Mining. Virtual Event, Israel: ACM, 2021: 958-966.

[12] Chen W, Hsu W, Lee ML. Making recommendations from multiple domains[C] //Proceedings of the 19th ACM SIGKDD International Conference on Knowledge Discovery and Data Mining. Chicago, IL, USA: ACM, 2013: 892-900.

[13] Gao S, Luo H, Chen D, et al. Cross—domain recommendation via cluster—level latent factor model[C] //Machine Learning and Knowledge Discovery in Databases: European Conference. Prague, Czech Republic: Springer, 2013: 161-176.

[14] Moreno O, Shapira B, Rokach L, et al. TALMUD: transfer learning for multiple domains[C] //Proceedings of the 21st ACM International Conference on Information and Knowledge Management. Maui, HI, USA: ACM, 2012: 425-434.

[15] Yuan F, Yao L, Benatallah B. DARec: deep domain adaptation for cross—domain recommendation via transferring rating patterns[ C] //Proceedings of the Twenty—Eighth International Joint Conference on Artificial Intelligence. Macao, China: International Joint Conferences on Artificial Intelligence Organization, 2019: 4227-4233.

[16] Ma H, Yang H, Lyu MR, et al. SoRec: social recommendation using probabilistic matrix factorization[C] //Proceedings of the 17th ACM Conference on Information and Knowledge Management. Napa Valley, CA, USA: ACM, 2008: 931-940.

[17] Xin X, Liu Z, Lin CY, et el. Cross—domain collaborative filtering with review text[C] //Proceedings of the Twenty—Fourth International Joint Conference on Artificial Intelligence. Buenos Aires, Argentina: AAAI Press, 2015: 1827-1834.

[18] Samra A, Frolov E, Vasilev A, et al. Cross—domain latent factors sharing via implicit matrix factorization[C] //Proceedings of the 18th ACM Conference on Recommender Systems. Bari, Italy: ACM, 2024: 309-317.

[19] Guo L, Tang L, Chen T, et al. DA—GCN: a domain—aware attentive graph convolution network for shared—account cross—domain sequential recommendation [ C] //Proceedings of the Thirtieth International Joint Conference on Artificial Intelligence. Montreal, Canada: International Joint Conferences on Artificial Intelligence Organization, 2021: 2483-2489.

[20] Wang X, He X, Nie L, et al. Item Silk Road: recommending items from information domains to social users[C] //Proceedings of the 40th International ACM SIGIR Conference on Research and Development in Information Retrieval. Shinjuku, Tokyo, Japan: ACM,2017: 185-194.

[21] Xu K, Xie Y, Chen L, et al. Expanding relationship for cross domain recommendation[C] //Proceedings of the 30th ACM International Conference on Information & Knowledge Management. Virtual Event, Queensland, Australia: ACM, 2021: 2251-2260.

[22] Gao C, Chen X, Feng F, et al. Cross—domain recommendation without sharing user—relevant data[C] //Proceedings of The Web Conference 2019. San Francisco, CA, USA: ACM, 2019: 491-502.

[23] Kaang S, Hwang J, Lee D, et al. Semi—supervised learning for cross—domain recommendation to cold—start users[C] //Proceedings of the 28th ACM International Conference on Information and Knowledge Management. Beijing, China: ACM, 2019: 1563-1572.

[24] Cao J, Li S, Yu B, et al. Towards Universal cross—domain recommendation[C] //Proceedings of the Sixteenth ACM International Conference on Web Search and Data Mining. Singapore: ACM,2023: 78-86.

[25] Zhao C, Li C, Fu C. Cross—domain recommendation via preference propagation Graph Net[C] //Proceedings of the 28th ACM International Conference on Information and Knowledge Management. Beijing, China: ACM, 2019.

[26] Man T, Shen H, Jin X, et al. Cross—domain recommendation: an embedding and mapping approach[C] //Proceedings of the Twenty—Sixth International Joint Conference on Artificial Intelligence. Melbourne, Australia: International Joint Conferences on Artificial Intelligence Organization,2017: 2464-2470.

[27] Bi Y, Song L, Yao M, et al. A heterogeneous information network based cross domain insurance recommendation system for cold start users [ C] //Proceedings of the 43rd International ACM SIGIR Conference on Research and Development in Information Retrieval. Virtual Event, China: ACM, 2020.

[28] Fu W, Peng Z, Wang S, et al. Deeply fusing reviews and contents for cold start users in cross—domain recommendation systems[C] //Proceedings of the Thirty—Third AAAI Conference on Artificial Intelligence. Honolulu, Hawaii, USA: AAAI Press, 2019: 94-101.

[29] Wang X, Peng Z, Wang S, et al. Cross—domain recommendation for cold—start users via neighborhood based feature mapping[C] //Database Systems for Advanced Applications: 23rd International Conference, DASFAA 2018. Gold Coast, QLD, Australia: Springer, 2018: 158-165.

[30] Li H, Ma W, Sun P, et al. Aiming at the target: filter collaborative information for cross—domain recommendation[ C] //Proceedings of the 47th International ACM SIGIR Conference on Research and Development in Information Retrieval. Washington, DC, USA: ACM, 2024: 2081-2085.

[31] Wang T, Zhuang F, Zhang Z, et al. Low—dimensional alignment for cross—domain recommendation[ C] //Proceedings of the 30th ACM International Conference on Information & Knowledge Management. Virtual Event, Queensland, Australia: ACM, 2021: 3508-3512.

[32] Zhu Y, Ge K, Zhuang F, et al. Transfer—meta framework for cross—domain recommendation to cold—start users [ C] //Proceedings of the 44th International ACM SIGIR Conference on Research and Development in Information Retrieval. Virtual Event, Canada: ACM, 2021: 1813-1817.

[33] Zhu Y, Tang Z, Liu Y, et al. Personalized transfer of user preferences for cross—domain recommendation[C] //Proceedings of the Fifteenth ACM International Conference on Web Search and Data Mining. Virtual Event, Tempe, AZ, USA: ACM,2022: 1507-1515.

[34] Chen C, Guo J, Song B. Dual attention transfer in session—based recommendation with multi—dimensional integration[C] //Proceedings of the 44th International ACM SIGIR Conference on Research and Development in Information Retrieval. Virtual Event, Canada: ACM, 2021: 869-878.

[35] Li P, Jiang Z, Que M, et al. Dual attentive sequential learning for cross—domain click—through rate prediction [ C] //Proceedings of the 27th ACM SIGKDD Conference on Knowledge Discovery & Data Mining. Virtual Event, Singapore: ACM, 2021: 3172-3180.

[36] Wang Y, Xia Y, Zhao L, et al. Dual transfer learning for neural machine translation with marginal distribution regularization[C] //Proceedings of the Thirty—Second AAAI Conference on Artificial Intelligence. New Orleans, LA, USA: AAAI Press, 2018: 5553-5560.

[37] 杜权, 曾信, 李北, 等. 面向标记语言的机器翻译方法研究[J].中文信息学报, 2024, 38(8): 68-75.

[38] Li P, Tuzhilin A. DDTCDR: deep dual transfer cross domain recommendation[ C] //Proceedings of the Thirteenth ACM International Conference on Web Search and Data Mining. Houston, TX, USA: ACM, 2020: 331-339.

[39] Zhao C, Li C, Xiao R, et al. CATN: cross—domain recommendation for cold—start users via aspect transfer network[ C] //Proceedings of the 43rd International ACM SIGIR Conference on Research and Development in Information Retrieval. Virtual Event, China: ACM, 2020: 229-238.

[40] Bengio Y, Courville A, Vincent P. Representation learning: a review and new perspectives[J].IEEE Transactions on Pattern Analysis and Machine Intelligence, 2013, 35(8): 1798-1828.

[41] 张佳, 张丽红. 基于条件增强和注意力机制的文本生成图像方法[J].测试技术学报, 2023, 37(2): 112-119.

[42] Eupont E. Learning disentangled joint continuous and discrete representations[EB/ OL].arXiv: 1804.00104, 2018.

[43] Gidaris S, Singh P, Komodakis N. Unsupervised representation learning by predicting image rotations[EB/ OL].arXiv: 1803.07728, 2018.

[44] Higgins I, Matthey L, Pal A, et al. beta—VAE: learning basic visual concepts with a constrained variational framework[C] //Proceedings of the 5th International Conference on Learning Representations. Toulon, France: OpenReview.net, 2017.

[45] Hsieh JT, Liu B, Huang DA, et al. Learning to decompose and disentangle representations for video prediction[EB/ OL]. arXiv: 1806.04166,2018.

[46] John V, Mou L, Bahuleyan H, et al. Disentangled representation learning for non—parallel text style transfer[C] //Proceedings of the 57th Annual Meeting of the Association for Computational Linguistics. Florence, Italy: Association for Computational Linguistics, 2019: 424-434.

[47] Li A, Cheng Z, Liu F, et al. Disentangled graph neural networks for session—based recommendation[J].IEEE Transactions on Knowledge and Data Engineering, 2023: 1-13.

[48] Ma J, Zhou C, Cui P, et al. Learning disentangled representations for recommendation[C] //Advances in Neural Information Processing Systems 32. Vancouver, Canada: Curran Associates, Inc., 2019: 5712-5723.

[49] Nema P, Karatzoglou A, Radlinski F. Disentangling preference representations for recommendation critiquing with β—VAE[C] //Proceedings of the 30th ACM International Conference on Information & Knowledge Management. Virtual Event, Queensland, Australia: ACM, 2021: 1356-1365.

[50] Wang X, Jin H, Zhang A, et al. Disentangled graph collaborative filtering[C] //Proceedings of the 43rd International ACM SIGIR Conference on Research and Development in Information Retrieval. Virtual Event, China: ACM, 2020: 1001-1010.

[51] Wang Y, Tang S, Lei Y, et al. DisenHAN: disentangled heterogeneous graph attention network for recommendation[ C] //Proceedings of the 29th ACM International Conference on Information & Knowledge Management. Virtual Event, Ireland: ACM, 2020: 1605-1614.

[52] Zheng Y, Gao C, Chang J, et al. Disentangling long and short—term interests for recommendation[C] //Proceedings of the ACM Web Conference 2022. Virtual Event, Lyon, France: ACM, 2022: 2256-2267.

[53] Zheng Y, Gao C, Li X, et al. Disentangling user interest and conformity for recommendation with causal embedding[C] //Proceedings of The Web Conference 2021. Virtual Event, Ljubljana, Slovenia: ACM, 2021: 2980-2991.

[54] Guo Z, Li G, Li J, et al. Dual VAE: Dual disentangled variational autoencoder for recommendation[C] //Proceedings of the 2024 SIAM International Conference on Data Mining. Houston, TX, USA: SIAM, 2024: 571-579.

[55] Zhang Y, Liu Y, Han P, et al. Learning personalized item set mapping for cross—domain recommendation[C] //Proceedings of the Twenty—Ninth International Joint Conference on Artificial Intelligence. Yokohama, Japan: International Joint Conferences on Artificial Intelligence Organization, 2020: 2561-2567.

[56] He X, Liao L, Zhang H, et al. Neural collaborative filtering[C] //Proceedings of the 26th International Conference on World Wide Web. Perth, Australia: International World Wide Web Conferences Steering Committee,2017: 173-182.

[57] Singh AP, Gordon GJ. Relational learning via collective matrix factorization[C] //Proceedings of the 14th ACM SIGKDD International Conference on Knowledge Discovery and Data Mining. Las Vegas, NV, USA: ACM, 2008: 650-658.

[58] Zhu F, Wang Y, Chen C, et al. A graphical and attentional framework for dual—target cross—domain recommendation[C] //Proceedings of the Twenty—Ninth International Joint Conference on Artificial Intelligence. Yokohama, Japan: International Joint Conferences on Artificial Intelligence Organization, 2020: 3001-3008.

[59] Zhao X, Yang N, Yu PS. Multi—sparse—domain collaborative recommendation via enhanced comprehensive aspect preference learning[C] //Proceedings of the Fifteenth ACM International Conference on Web Search and Data Mining. Virtual Event, Tempe, AZ, USA: ACM, 2022: 1452-1460.

[60] Ma M, Ren P, Lin Y, et al. π—Net: a parallel information—sharing network for shared—account cross—domain sequential recommendations[C] //Proceedings of the 42nd International ACM SIGIR Conference on Research and Development in Information Retrieval. Paris, France: ACM,2019: 685-694.

[61] Perera D, Zimmermann R. LSTM networks for online cross—network recommendations[C] //Proceedings of the Twenty—Seventh International Joint Conference on Artificial Intelligence. Stockholm, Sweden: International Joint Conferences on Artificial Intelligence Organization, 2018: 3825-3833.

基金资助

国家自然科学青年科学基金资助项目(62402215)

中车重大资助项目(2025CKA509)