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

General Debiasing for Graph-based Collaborative Filtering via Adversarial Graph Dropout (2402.13769v1)

Published 21 Feb 2024 in cs.IR

Abstract: Graph neural networks (GNNs) have shown impressive performance in recommender systems, particularly in collaborative filtering (CF). The key lies in aggregating neighborhood information on a user-item interaction graph to enhance user/item representations. However, we have discovered that this aggregation mechanism comes with a drawback, which amplifies biases present in the interaction graph. For instance, a user's interactions with items can be driven by both unbiased true interest and various biased factors like item popularity or exposure. However, the current aggregation approach combines all information, both biased and unbiased, leading to biased representation learning. Consequently, graph-based recommenders can learn distorted views of users/items, hindering the modeling of their true preferences and generalizations. To address this issue, we introduce a novel framework called Adversarial Graph Dropout (AdvDrop). It differentiates between unbiased and biased interactions, enabling unbiased representation learning. For each user/item, AdvDrop employs adversarial learning to split the neighborhood into two views: one with bias-mitigated interactions and the other with bias-aware interactions. After view-specific aggregation, AdvDrop ensures that the bias-mitigated and bias-aware representations remain invariant, shielding them from the influence of bias. We validate AdvDrop's effectiveness on five public datasets that cover both general and specific biases, demonstrating significant improvements. Furthermore, our method exhibits meaningful separation of subgraphs and achieves unbiased representations for graph-based CF models, as revealed by in-depth analysis. Our code is publicly available at https://github.com/Arthurma71/AdvDrop.

Definition Search Book Streamline Icon: https://streamlinehq.com
References (74)
  1. Controlling popularity bias in learning-to-rank recommendation. In Proceedings of the eleventh ACM conference on recommender systems. 42–46.
  2. The Unfairness of Popularity Bias in Recommendation. In RMSE@RecSys (CEUR Workshop Proceedings, Vol. 2440).
  3. Invariant Risk Minimization. CoRR abs/1907.02893 (2019).
  4. Privacy-aware recommendation with private-attribute protection using adversarial learning. In Proceedings of the 13th International Conference on Web Search and Data Mining. 34–42.
  5. Stephen Bonner and Flavian Vasile. 2018. Causal embeddings for recommendation. In Proceedings of the 12th ACM conference on recommender systems. 104–112.
  6. Avishek Bose and William Hamilton. 2019a. Compositional fairness constraints for graph embeddings. In International Conference on Machine Learning. PMLR, 715–724.
  7. Avishek Joey Bose and William L. Hamilton. 2019b. Compositional Fairness Constraints for Graph Embeddings. In ICML.
  8. Counterfactual Reasoning and Learning Systems: The Example of Computational Advertising. Journal of Machine Learning Research 14, 11 (2013).
  9. Learning to rank: from pairwise approach to listwise approach. In ICML, Vol. 227. 129–136.
  10. A probabilistic model for using social networks in personalized item recommendation. In Proceedings of the 9th ACM Conference on Recommender Systems. 43–50.
  11. AutoDebias: Learning to Debias for Recommendation. In SIGIR.
  12. Bias and Debias in Recommender System: A Survey and Future Directions. ACM Trans. Inf. Syst. 41, 3 (2023), 67:1–67:39.
  13. Revisiting Graph Based Collaborative Filtering: A Linear Residual Graph Convolutional Network Approach. In AAAI.
  14. A Simple Framework for Contrastive Learning of Visual Representations. In ICML. 1597–1607.
  15. Co-training Disentangled Domain Adaptation Network for Leveraging Popularity Bias in Recommenders. In SIGIR.
  16. Co-training Disentangled Domain Adaptation Network for Leveraging Popularity Bias in Recommenders. In Proceedings of the 45th International ACM SIGIR Conference on Research and Development in Information Retrieval. 60–69.
  17. Esam: Discriminative domain adaptation with non-displayed items to improve long-tail performance. In Proceedings of the 43rd International ACM SIGIR Conference on Research and Development in Information Retrieval. 579–588.
  18. Environment Inference for Invariant Learning. In ICML, Vol. 139. 2189–2200.
  19. Benchmarking Graph Neural Networks. JMLR 24 (2023), 43:1–43:48.
  20. A fairness-aware hybrid recommender system. arXiv preprint arXiv:1809.09030 (2018).
  21. KuaiRec: A Fully-Observed Dataset and Insights for Evaluating Recommender Systems. In CIKM. 540–550.
  22. Fairness-aware ranking in search & recommendation systems with application to linkedin talent search. In Proceedings of the 25th acm sigkdd international conference on knowledge discovery & data mining. 2221–2231.
  23. Offline Evaluation to Make Decisions About Playlist Recommendation Algorithms. In WSDM.
  24. Offline evaluation to make decisions about playlistrecommendation algorithms. In Proceedings of the Twelfth ACM International Conference on Web Search and Data Mining. 420–428.
  25. Inductive Representation Learning on Large Graphs. In NIPS.
  26. LightGCN: Simplifying and Powering Graph Convolution Network for Recommendation. In SIGIR.
  27. Unbiased learning-to-rank with biased feedback. In Proceedings of the tenth ACM international conference on web search and data mining. 781–789.
  28. Toshihiro Kamishima and Shotaro Akaho. 2017. Considerations on recommendation independence for a find-good-items task. (2017).
  29. Out-of-Distribution Generalization via Risk Extrapolation (REx). In ICML, Vol. 139. 5815–5826.
  30. Improving one-class collaborative filtering by incorporating rich user information. In Proceedings of the 19th ACM international conference on Information and knowledge management. 959–968.
  31. R-Drop: Regularized Dropout for Neural Networks. In NeurIPS. 10890–10905.
  32. Are you influenced by others when rating? Improve rating prediction by conformity modeling. In Proceedings of the 10th ACM conference on recommender systems. 269–272.
  33. Learning to recommend with social trust ensemble. In Proceedings of the 32nd international ACM SIGIR conference on Research and development in information retrieval. 203–210.
  34. UltraGCN: Ultra Simplification of Graph Convolutional Networks for Recommendation. In CIKM.
  35. Benjamin M. Marlin and Richard S. Zemel. 2009. Collaborative prediction and ranking with non-random missing data. In RecSys.
  36. A Survey on Bias and Fairness in Machine Learning. ACM Comput. Surv. 54, 6 (2022), 115:1–115:35.
  37. Exploring Data Splitting Strategies for the Evaluation of Recommendation Models. In RecSys. ACM, 681–686.
  38. Fairwalk: Towards fair graph embedding. (2019).
  39. Steffen Rendle. 2022. Item Recommendation from Implicit Feedback. In Recommender Systems Handbook. Springer US, 143–171.
  40. BPR: Bayesian Personalized Ranking from Implicit Feedback. CoRR abs/1205.2618.
  41. Unbiased Recommender Learning from Missing-Not-At-Random Implicit Feedback. In WSDM. 501–509.
  42. Recommendations as treatments: Debiasing learning and evaluation. In international conference on machine learning. PMLR, 1670–1679.
  43. Deep Crossing: Web-Scale Modeling without Manually Crafted Combinatorial Features. In KDD. ACM, 255–262.
  44. Ashudeep Singh and Thorsten Joachims. 2018. Fairness of exposure in rankings. In Proceedings of the 24th ACM SIGKDD International Conference on Knowledge Discovery & Data Mining. 2219–2228.
  45. Session-Based Social Recommendation via Dynamic Graph Attention Networks. In WSDM. ACM, 555–563.
  46. Laurens Van der Maaten and Geoffrey Hinton. 2008. Visualizing data using t-SNE. Journal of machine learning research 9, 11 (2008).
  47. Click” is not equal to “like”: Counterfactual recommendation for mitigating clickbait issue. arXiv preprint arXiv:2009.09945 (2020).
  48. Clicks can be cheating: Counterfactual recommendation for mitigating clickbait issue. In Proceedings of the 44th International ACM SIGIR Conference on Research and Development in Information Retrieval. 1288–1297.
  49. Causal Representation Learning for Out-of-Distribution Recommendation. In WWW.
  50. Neural Graph Collaborative Filtering. In SIGIR.
  51. Learning personalized preference of strong and weak ties for social recommendation. In Proceedings of the 26th International Conference on World Wide Web. 1601–1610.
  52. Multi-Component Graph Convolutional Collaborative Filtering. In AAAI.
  53. Doubly robust joint learning for recommendation on data missing not at random. In International Conference on Machine Learning. PMLR, 6638–6647.
  54. Information Theoretic Counterfactual Learning from Missing-Not-At-Random Feedback. In NeurIPS.
  55. Invariant Preference Learning for General Debiasing in Recommendation. In KDD.
  56. Model-agnostic counterfactual reasoning for eliminating popularity bias in recommender system. In Proceedings of the 27th ACM SIGKDD Conference on Knowledge Discovery & Data Mining. 1791–1800.
  57. Distributionally-robust Recommendations for Improving Worst-case User Experience. In WWW.
  58. Self-supervised Graph Learning for Recommendation. In SIGIR.
  59. Deconfounded causal collaborative filtering. arXiv preprint arXiv:2110.07122 (2021).
  60. Top-n recommendation with counterfactual user preference simulation. In Proceedings of the 30th ACM International Conference on Information & Knowledge Management. 2342–2351.
  61. Sirui Yao and Bert Huang. 2017a. Beyond parity: Fairness objectives for collaborative filtering. Advances in neural information processing systems 30 (2017).
  62. Sirui Yao and Bert Huang. 2017b. New fairness metrics for recommendation that embrace differences. arXiv preprint arXiv:1706.09838 (2017).
  63. Mingzhang Yin and Mingyuan Zhou. 2019. ARM: Augment-REINFORCE-Merge Gradient for Stochastic Binary Networks. In ICLR (Poster). OpenReview.net.
  64. Selection of negative samples for one-class matrix factorization. In Proceedings of the 2017 SIAM International Conference on Data Mining. SIAM, 363–371.
  65. Incorporating Bias-aware Margins into Contrastive Loss for Collaborative Filtering. arXiv preprint arXiv:2210.11054 (2022).
  66. Robust Collaborative Filtering to Popularity Distribution Shift. ACM Transactions on Information Systems (2023).
  67. Empowering Collaborative Filtering with Principled Adversarial Contrastive Loss. In NeurIPS.
  68. Invariant Collaborative Filtering to Popularity Distribution Shift. In WWW.
  69. Investigating Accuracy-Novelty Performance for Graph-based Collaborative Filtering. (2022).
  70. Popularity bias is not always evil: Disentangling benign and harmful bias for recommendation. IEEE Transactions on Knowledge and Data Engineering (2022).
  71. Disentangling User Interest and Conformity for Recommendation with Causal Embedding. In WWW. 2980–2991.
  72. Adaptive Popularity Debiasing Aggregator for Graph Collaborative Filtering. In SIGIR.
  73. Popularity-Opportunity Bias in Collaborative Filtering. In WSDM. ACM, 85–93.
  74. Measuring and mitigating item under-recommendation bias in personalized ranking systems. In Proceedings of the 43rd international ACM SIGIR conference on research and development in information retrieval. 449–458.
User Edit Pencil Streamline Icon: https://streamlinehq.com
Authors (5)
  1. An Zhang (78 papers)
  2. Wenchang Ma (8 papers)
  3. Pengbo Wei (4 papers)
  4. Leheng Sheng (9 papers)
  5. Xiang Wang (279 papers)
Citations (3)
View on Semantic Scholar

Summary

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

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