Papers
Topics
Authors
Recent
Gemini 2.5 Flash
Gemini 2.5 Flash
125 tokens/sec
GPT-4o
53 tokens/sec
Gemini 2.5 Pro Pro
42 tokens/sec
o3 Pro
4 tokens/sec
GPT-4.1 Pro
47 tokens/sec
DeepSeek R1 via Azure Pro
28 tokens/sec
2000 character limit reached

Causal Deconfounding via Confounder Disentanglement for Dual-Target Cross-Domain Recommendation (2404.11180v3)

Published 17 Apr 2024 in cs.IR

Abstract: In recent years, dual-target Cross-Domain Recommendation (CDR) has been proposed to capture comprehensive user preferences in order to ultimately enhance the recommendation accuracy in both data-richer and data-sparser domains simultaneously. However, in addition to users' true preferences, the user-item interactions might also be affected by confounders (e.g., free shipping, sales promotion). As a result, dual-target CDR has to meet two challenges: (1) how to effectively decouple observed confounders, including single-domain confounders and cross-domain confounders, and (2) how to preserve the positive effects of observed confounders on predicted interactions, while eliminating their negative effects on capturing comprehensive user preferences. To address the above two challenges, we propose a Causal Deconfounding framework via Confounder Disentanglement for dual-target Cross-Domain Recommendation, called CD2CDR. In CD2CDR, we first propose a confounder disentanglement module to effectively decouple observed single-domain and cross-domain confounders. We then propose a causal deconfounding module to preserve the positive effects of such observed confounders and eliminate their negative effects via backdoor adjustment, thereby enhancing the recommendation accuracy in each domain. Extensive experiments conducted on five real-world datasets demonstrate that CD2CDR significantly outperforms the state-of-the-art methods.

Definition Search Book Streamline Icon: https://streamlinehq.com
References (58)
  1. F. Zhu, C. Chen, Y. Wang, G. Liu, and X. Zheng, “Dtcdr: A framework for dual-target cross-domain recommendation,” in CIKM, 2019, pp. 1533–1542.
  2. C. Gao, Y. Zheng, W. Wang, F. Feng, X. He, and Y. Li, “Causal inference in recommender systems: A survey and future directions,” ACM TOIS, 2022.
  3. Y. Zhu, J. Ma, and J. Li, “Causal inference in recommender systems: A survey of strategies for bias mitigation, explanation, and generalization,” arXiv preprint arXiv:2301.00910, 2023.
  4. A. Wu, K. Kuang, R. Xiong, and F. Wu, “Instrumental variables in causal inference and machine learning: A survey,” arXiv preprint arXiv:2212.05778, 2022.
  5. H. Luo, F. Zhuang, R. Xie, H. Zhu, D. Wang, Z. An, and Y. Xu, “A survey on causal inference for recommendation,” The Innovation, 2024.
  6. S. C. Gowda, S. Joshi, H. Zhang, and M. Ghassemi, “Pulling up by the causal bootstraps: Causal data augmentation for pre-training debiasing,” in CIKM, 2021, pp. 606–616.
  7. X. Zhu, Y. Zhang, X. Yang, D. Wang, and F. Feng, “Mitigating hidden confounding effects for causal recommendation,” IEEE TKDE, 2024.
  8. Y. Zhang, F. Feng, X. He, T. Wei, C. Song, G. Ling, and Y. Zhang, “Causal intervention for leveraging popularity bias in recommendation,” in SIGIR, 2021, pp. 11–20.
  9. W. Wang, F. Feng, X. He, X. Wang, and T.-S. Chua, “Deconfounded recommendation for alleviating bias amplification,” in KDD, 2021, pp. 1717–1725.
  10. S. Li, L. Yao, S. Mu, W. X. Zhao, Y. Li, T. Guo, B. Ding, and J.-R. Wen, “Debiasing learning based cross-domain recommendation,” in KDD, 2021, pp. 3190–3199.
  11. Y. Zhang, Y. Shen, D. Wang, J. Gu, and G. Zhang, “Connecting unseen domains: Cross-domain invariant learning in recommendation,” in SIGIR, 2023, pp. 1894–1898.
  12. Q. Li, X. Wang, Z. Wang, and G. Xu, “Be causal: De-biasing social network confounding in recommendation,” ACM TKDD, pp. 1–23, 2023.
  13. D. Yu, Q. Li, X. Wang, and G. Xu, “Deconfounded recommendation via causal intervention,” Neurocomputing, pp. 128–139, 2023.
  14. X. Du, Z. Wu, F. Feng, X. He, and J. Tang, “Invariant representation learning for multimedia recommendation,” in MM, 2022, pp. 619–628.
  15. Z. Li, D. Amagata, Y. Zhang, T. Hara, S. Haruta, K. Yonekawa, and M. Kurokawa, “Debiasing graph transfer learning via item semantic clustering for cross-domain recommendations,” in Big Data, 2022, pp. 762–769.
  16. J. Cao, J. Sheng, X. Cong, T. Liu, and B. Wang, “Cross-domain recommendation to cold-start users via variational information bottleneck,” in ICDE, 2022, pp. 2209–2223.
  17. Z. Wang, S. Shen, Z. Wang, B. Chen, X. Chen, and J.-R. Wen, “Unbiased sequential recommendation with latent confounders,” in WWW, 2022, pp. 2195–2204.
  18. S. Xu, J. Tan, S. Heinecke, V. J. Li, and Y. Zhang, “Deconfounded causal collaborative filtering,” ACM TORS, pp. 1–25, 2023.
  19. R. Zhan, C. Pei, Q. Su, J. Wen, X. Wang, G. Mu, D. Zheng, P. Jiang, and K. Gai, “Deconfounding duration bias in watch-time prediction for video recommendation,” in KDD, 2022, pp. 4472–4481.
  20. F. Zhu, Y. Wang, C. Chen, J. Zhou, L. Li, and G. Liu, “Cross-domain recommendation: Challenges, progress, and prospects,” in IJCAI, 2021, pp. 4721–4728.
  21. H. Kanagawa, H. Kobayashi, N. Shimizu, Y. Tagami, and T. Suzuki, “Cross-domain recommendation via deep domain adaptation,” in ECIR, 2019, pp. 20–29.
  22. G. Hu, Y. Zhang, and Q. Yang, “Transfer meets hybrid: A synthetic approach for cross-domain collaborative filtering with text,” in WWW, 2019, pp. 2822–2829.
  23. W. Fu, Z. Peng, S. Wang, Y. Xu, and J. Li, “Deeply fusing reviews and contents for cold start users in cross-domain recommendation systems,” in AAAI, 2019, pp. 94–101.
  24. F. Yuan, L. Yao, and B. Benatallah, “Darec: Deep domain adaptation for cross-domain recommendation via transferring rating patterns,” in IJCAI, 2019, pp. 4227–4233.
  25. F. Zhu, Y. Wang, J. Zhou, C. Chen, L. Li, and G. Liu, “A unified framework for cross-domain and cross-system recommendations,” IEEE TKDE, 2021.
  26. X. Guo, S. Li, N. Guo, J. Cao, X. Liu, Q. Ma, R. Gan, and Y. Zhao, “Disentangled representations learning for multi-target cross-domain recommendation,” ACM TOIS, pp. 1–27, 2023.
  27. P. Li and A. Tuzhilin, “Ddtcdr: Deep dual transfer cross domain recommendation,” in WSDM, 2020, pp. 331–339.
  28. M. Liu, J. Li, G. Li, and P. Pan, “Cross domain recommendation via bi-directional transfer graph collaborative filtering networks,” in CIKM, 2020, pp. 885–894.
  29. J. Cao, X. Lin, X. Cong, J. Ya, T. Liu, and B. Wang, “Disencdr: Learning disentangled representations for cross-domain recommendation,” in SIGIR, 2022, pp. 267–277.
  30. R. Zhang, T. Zang, Y. Zhu, C. Wang, K. Wang, and J. Yu, “Disentangled contrastive learning for cross-domain recommendation,” in DASFAA, 2023, pp. 163–178.
  31. P. Wu, H. Li, Y. Deng, W. Hu, Q. Dai, Z. Dong, J. Sun, R. Zhang, and X.-H. Zhou, “On the opportunity of causal learning in recommendation systems: Foundation, estimation, prediction and challenges,” in IJCAI, 2022, pp. 5646–5653.
  32. S. Xu, J. Ji, Y. Li, Y. Ge, J. Tan, and Y. Zhang, “Causal inference for recommendation: Foundations, methods and applications,” arXiv preprint arXiv:2301.04016, 2023.
  33. M. Sato, S. Takemori, J. Singh, and T. Ohkuma, “Unbiased learning for the causal effect of recommendation,” in RecSys, 2020, pp. 378–387.
  34. D. Liu, P. Cheng, H. Zhu, Z. Dong, X. He, W. Pan, and Z. Ming, “Mitigating confounding bias in recommendation via information bottleneck,” in RecSys, 2021, pp. 351–360.
  35. ——, “Debiased representation learning in recommendation via information bottleneck,” ACM TORS, pp. 1–27, 2023.
  36. Y. Wang, D. Liang, L. Charlin, and D. M. Blei, “Causal inference for recommender systems,” in RecSys, 2020, pp. 426–431.
  37. Q. Zhang, X. Zhang, Y. Liu, H. Wang, M. Gao, J. Zhang, and R. Guo, “Debiasing recommendation by learning identifiable latent confounders,” in KDD, 2023, p. 3353–3363.
  38. W. Xu, Q. Wu, R. Wang, M. Ha, Q. Ma, L. Chen, B. Han, and J. Yan, “Rethinking cross-domain sequential recommendation under open-world assumptions,” arXiv preprint arXiv:2311.04590, 2023.
  39. C. Zhao, X. Li, M. He, H. Zhao, and J. Fan, “Sequential recommendation via an adaptive cross-domain knowledge decomposition,” in CIKM, 2023, pp. 3453–3463.
  40. J. Ma, C. Zhou, P. Cui, H. Yang, and W. Zhu, “Learning disentangled representations for recommendation,” in NeurIPS, 2019, pp. 5711–5722.
  41. Y. Zheng, C. Gao, X. Li, X. He, Y. Li, and D. Jin, “Disentangling user interest and conformity for recommendation with causal embedding,” in WWW, 2021, pp. 2980–2991.
  42. X. Wang, Q. Li, D. Yu, P. Cui, Z. Wang, and G. Xu, “Causal disentanglement for semantics-aware intent learning in recommendation,” IEEE TKDE, 2022.
  43. J. Zhu, Y. Wang, F. Zhu, and Z. Sun, “Domain disentanglement with interpolative data augmentation for dual-target cross-domain recommendation,” in RecSys, 2023, pp. 515–527.
  44. B. Schölkopf, D. W. Hogg, D. Wang, D. Foreman-Mackey, D. Janzing, C.-J. Simon-Gabriel, and J. Peters, “Modeling confounding by half-sibling regression,” PNAS, pp. 7391–7398, 2016.
  45. J. D. M.-W. C. Kenton and L. K. Toutanova, “Bert: Pre-training of deep bidirectional transformers for language understanding,” in NAACL-HLT, 2019, pp. 4171–4186.
  46. J.-Y. Zhu, T. Park, P. Isola, and A. A. Efros, “Unpaired image-to-image translation using cycle-consistent adversarial networks,” in ICCV, 2017, pp. 2223–2232.
  47. S. Zhang, X. Feng, W. Fan, W. Fang, F. Feng, W. Ji, S. Li, L. Wang, S. Zhao, Z. Zhao et al., “Video-audio domain generalization via confounder disentanglement,” in AAAI, 2023, pp. 15 322–15 330.
  48. H. Su, J. Li, Z. Du, L. Zhu, K. Lu, and H. T. Shen, “Cross-domain recommendation via dual adversarial adaptation,” ACM TOIS, 2023.
  49. I. Goodfellow, J. Pouget-Abadie, M. Mirza, B. Xu, D. Warde-Farley, S. Ozair, A. Courville, and Y. Bengio, “Generative adversarial nets,” NeurIPS, 2014.
  50. Y. Zhu, X. Ren, J. Yi, and Z. Chen, “Deep deconfounded content-based tag recommendation for ugc with causal intervention,” arXiv preprint arXiv:2205.14380, 2022.
  51. X. He, L. Liao, H. Zhang, L. Nie, X. Hu, and T.-S. Chua, “Neural collaborative filtering,” in WWW, 2017, pp. 173–182.
  52. F. Zhu, Y. Wang, C. Chen, G. Liu, and X. Zheng, “A graphical and attentional framework for dual-target cross-domain recommendation,” in IJCAI, 2020, pp. 3001–3008.
  53. D. P. Kingma and J. Ba, “Adam: A method for stochastic optimization,” in ICLR, 2015.
  54. X. Wang, X. He, M. Wang, F. Feng, and T.-S. Chua, “Neural graph collaborative filtering,” in SIGIR, 2019, pp. 165–174.
  55. X. He, K. Deng, X. Wang, Y. Li, Y. Zhang, and M. Wang, “Lightgcn: Simplifying and powering graph convolution network for recommendation,” in SIGIR, 2020, pp. 639–648.
  56. T. Man, H. Shen, X. Jin, and X. Cheng, “Cross-domain recommendation: An embedding and mapping approach,” in IJCAI, 2017, pp. 2464–2470.
  57. F. Zhu, Y. Wang, C. Chen, G. Liu, M. Orgun, and J. Wu, “A deep framework for cross-domain and cross-system recommendations,” in IJCAI, 2018, pp. 3711–3717.
  58. Z. Lin, H. Ding, N. T. Hoang, B. Kveton, A. Deoras, and H. Wang, “Pre-trained recommender systems: A causal debiasing perspective,” in WSDM, 2024, pp. 424–433.
Citations (1)
View on Semantic Scholar

Summary

We haven't generated a summary for this paper yet.

Ai Sparkles Streamline Icon: https://streamlinehq.com Summarize Now