Bias and Debias in Recommender System: A Survey and Future Directions

By Jiawei Chen, Hande Dong, Xiang Wang, Fuli Feng, Meng Wang, Xiangnan He. ACM Trans. Inf. Syst., 2020

923 citations recorded by the source corpus.

While recent years have witnessed a rapid growth of research papers on recommender system (RS), most of the papers focus on inventing machine learning models to better fit user behavior data. However, user behavior data is observational rather than experimental. This makes various biases widely exist in the data, including but not limited to selection bias, position bias, exposure bias, and popularity bias. Blindly fitting the data without considering the inherent biases will result in many serious issues, e.g., the discrepancy between offline evaluation and online metrics, hurting user satisfaction and trust on the recommendation service, and so on. To transform the large volume of research models into practical improvements, it is highly urgent to explore the impacts of the biases and perform debiasing when necessary. When reviewing the papers that consider biases in RS, we find that, to our surprise, the studies are rather fragmented and lack a systematic organization. The terminology “bias” is widely used in the literature, but its definition is usually vague and even inconsistent across papers. This motivates us to provide a systematic survey of existing work on RS biases. In this paper, we first summarize seven types of biases in recommendation, along with their definitions and characteristics. We then provide a taxonomy to position and organize the existing work on recommendation debiasing. Finally, we identify some open challenges and envision some future directions, with the hope of inspiring more research work on this important yet less investigated topic. The summary of debiasing methods reviewed in this survey can be found at https://github.com/jiawei-chen/RecDebiasing.

Subfields: Computer Science Applications, Information Systems, Management Science and Operations Research

View Paper: https://www.semanticscholar.org/paper/9779f919685adee936835eff46914d3d69d9ccb9

How researchers see this paper

The researchers whose trusted networks rank it highest, strongest advocate first. Every researcher gets a different order.

In Computer Science Applications

In Information Systems

In Management Science and Operations Research

Citations in the corpus

Recommender Systems in the Era of Large Language Models (LLMs) (2023, neutral): “In this section, we briefly review some related work on recommender systems and LLMs techniques.”
Self-supervised Graph Learning for Recommendation (2020, background): “Hence, the GCNs are easily biased towards high-degree items [5, 36], sacrificing the performance of low-degree (long-tail) items.”
A Survey on Accuracy-Oriented Neural Recommendation: From Collaborative Filtering to Information-Rich Recommendation (2021, methodology): “The data-driven approaches with accuracy as goals may lead to biases in the algorithmic process decision process [266], [267], [268], [269].”
Self-Supervised Learning for Recommender Systems: A Survey (2022, methodology): “Similarity prediction is also used to address selection bias in recommender systems [39].”
XSimGCL: Towards Extremely Simple Graph Contrastive Learning for Recommendation (2022, background): “Optimizing this contrastive loss always leads to more evenly distributed user/item representations regardless of graph augmentations, which implicitly alleviates the pervasive popularity bias [23] and promotes long-tail items.”
Graph Neural Networks in Recommender Systems: A Survey (2020, background): “There exist surveys focusing on different perspectives of recommender systems [4, 16, 22, 28, 44, 116, 199].”
Advances and Challenges in Conversational Recommender Systems: A Survey (2021, positive): “are survey papers focusing on certain cutting-edge aspects in recommender systems, such as the bias issues and debiasing methods [24], explainability/interpretability [249], evaluation issues [179], and novel methods that leverage deep neural networks [247, 225, 226], knowledge graphs [61], or reinf”
Causal Intervention for Leveraging Popularity Bias in Recommendation (2021, background): “Some other works have attempted to remove the impact of other bias problems [11], position bias [5], selection bias [43], and exposure bias [12, 26] etc.”
Model-Agnostic Counterfactual Reasoning for Eliminating Popularity Bias in Recommender System (2020, methodology): “We refer the readers to a systemic survey for more details [11].”
Are Graph Augmentations Necessary?: Simple Graph Contrastive Learning for Recommendation (2021, background): “Optimizing the contrastive loss InfoNCE [18] learns more uniform user/item representations no matter if graph augmentations are applied, which implicitly plays a role in mitigating the popularity bias [4].”
Learning Intents behind Interactions with Knowledge Graph for Recommendation (2021, background): “Furthermore, wewould like to introduce causal concepts, such as causal effect inference, counterfactual reasoning, and deconfounding, into knowledgeaware recommendation to discover and amplify biases [6].”
Disentangling User Interest and Conformity for Recommendation with Causal Embedding (2021, background): “Disentangled representation learning in recommender systems was not explored until recently [14, 33, 47].”
Large Language Models are Zero-Shot Rankers for Recommender Systems (2023, background): “The biases and debiasing methods in conventional recommender systems have been widely studied [2].”
Denoising Implicit Feedback for Recommendation (2020, background): “This is because implicit interactions are easily affected by the first impression of users and other factors [4, 37] such as caption bias [18] and position bias [20].”
A Survey on the Fairness of Recommender Systems (2022, background): “[16] recently reviewed bias in recommender systems and introduced fairness issues, but fairness is not their main focus, and”