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

Chen, Jiawei; Dong, Hande; Wang, Xiang; Feng, Fuli; Wang, Meng; He, Xiangnan

doi:10.48550/arxiv.2010.03240

Cited by 99 publications

(193 citation statements)

References 123 publications

Supporting

Mentioning

173

Contrasting

Order By: Relevance

“…We attribute improvements to several factors of STOSA: (1). the distribution representations help expand the latent interaction space of items to better understand uncertainty and flexibility; (2). the consideration of collaborative transitivity enhances the discovery and induction of collaborative signals inherent in item-item transitions; (4).…”

Section: Overall Comparison (Rq1 and Rq2)mentioning

confidence: 99%

“…Therefore, the overall Wasserstein self-attention time complexity is 𝑂 (𝑛𝑑 + 𝑛 2 𝑑 + 4𝑛 2 ). By also considering the feedforward networks, we obtain the final asymptotic computational complexity as 𝑂 (𝑛𝑑 +𝑛 2 𝑑 + 4𝑛 2 + 𝑛𝑑 2 2 ). The computation complexity of traditional self-attention [17] is 𝑂 (𝑛 2 𝑑 + 𝑛𝑑 2 ).…”

Section: Appendices a Complexity Analysismentioning

confidence: 99%

“…The incorporation of the distance between positive items and negatively sampled items is reasonable and necessary. Typically, negative items are sampled from items that the user never shows interest with, even if randomly sampled negative items are not necessarily hard negative items [47] due to data bias [2]. A proper way to sample negative items is an important topic in the recommendation, but it is beyond this paper's scope.…”

Section: Introductionmentioning

confidence: 99%

See 2 more Smart Citations

Sequential Recommendation via Stochastic Self-Attention

Fan¹,

Liu²,

Wang³

et al. 2022

Preprint

View full text Add to dashboard Cite

Sequential recommendation models the dynamics of a user's previous behaviors in order to forecast the next item, and has drawn a lot of attention. Transformer-based approaches, which embed items as vectors and use dot-product self-attention to measure the relationship between items, demonstrate superior capabilities among existing sequential methods. However, users' real-world sequential behaviors are uncertain rather than deterministic, posing a significant challenge to present techniques. We further suggest that dot-product-based approaches cannot fully capture collaborative transitivity, which can be derived in item-item transitions inside sequences and is beneficial for cold start items. We further argue that BPR loss has no constraint on positive and sampled negative items, which misleads the optimization.We propose a novel STOchastic Self-Attention (STOSA) to overcome these issues. STOSA, in particular, embeds each item as a stochastic Gaussian distribution, the covariance of which encodes the uncertainty. We devise a novel Wasserstein Self-Attention module to characterize item-item position-wise relationships in sequences, which effectively incorporates uncertainty into model training. Wasserstein attentions also enlighten the collaborative transitivity learning as it satisfies triangle inequality. Moreover, we introduce a novel regularization term to the ranking loss, which assures the dissimilarity between positive and the negative items. Extensive experiments on five real-world benchmark datasets demonstrate the superiority of the proposed model over state-of-the-art baselines, especially on cold start items. The code is available in https://github.com/zfan20/STOSA.

show abstract

Section: Overall Comparison (Rq1 and Rq2)mentioning

confidence: 99%

Section: Appendices a Complexity Analysismentioning

confidence: 99%

Section: Introductionmentioning

confidence: 99%

See 1 more Smart Citation

Sequential Recommendation via Stochastic Self-Attention

Fan¹,

Liu²,

Wang³

et al. 2022

Preprint

View full text Add to dashboard Cite

show abstract

Section: Fairness/bias In Recommendation Systemsmentioning

confidence: 99%

“…Recommendation Systems (RS) provide users with personalized suggestions and can help alleviate information overload [8]. While much recent work in RS investigates improved machine learning models for recommendation [8], recent years have seen a rise in the number of works examining fairness and bias in recommendation. In brief, unfairness in recommendations manifests as systematic discrimination against certain individuals in favor of others [15] based on protected attributes such as gender and age.…”

Section: Fairness/bias In Recommendation Systemsmentioning

confidence: 99%

Unintended Bias in Language Model-driven Conversational Recommendation

Shen¹,

Li²,

Bouadjenek³

et al. 2022

Preprint

View full text Add to dashboard Cite

Conversational Recommendation Systems (CRSs) have recently started to leverage pretrained language models (LM) such as BERT for their ability to semantically interpret a wide range of preference statement variations. However, pretrained LMs are well-known to be prone to intrinsic biases in their training data, which may be exacerbated by biases embedded in domain-specific language data (e.g., user reviews) used to fine-tune LMs for CRSs. We study a recently introduced LM-driven recommendation backbone (termed LMRec) of a CRS to investigate how unintended bias -i.e., language variations such as name references or indirect indicators of sexual orientation or location that should not affect recommendations -manifests in significantly shifted price and category distributions of restaurant recommendations. The alarming results we observe strongly indicate that LMRec has learned to reinforce harmful stereotypes through its recommendations. For example, offhand mention of names associated with the black community significantly lowers the price distribution of recommended restaurants, while offhand mentions of common male-associated names lead to an increase in recommended alcohol-serving establishments. These and many related results presented in this work raise a red flag that advances in the language handling capability of LM-driven CRSs do not come without significant challenges related to mitigating unintended bias in future deployed CRS assistants with a potential reach of hundreds of millions of end users.

show abstract

Recommender systems, ground truth, and preference pollution

et al. 2022

View full text Add to dashboard Cite

Interactions between individuals and recommender systems can be viewed as a continuous feedback loop, consisting of pre-consumption and post-consumption phases. Pre-consumption, systems provide recommendations that are typically based on predictions of user preferences. They represent a valuable service for both providers and users as decision aids. After item consumption, the user provides post-consumption feedback (e.g., a preference rating) to the system, often used to improve the system's subsequent recommendations, completing the feedback loop. There is a growing understanding that this feedback loop can be a significant source of unintended consequences, introducing decisionmaking biases that can affect the quality of the "ground truth" preference data, which serves as the key input to modern recommender systems. This paper highlights two forms of bias that recommender systems inherently inflict on the "ground truth" preference data collected from users after item consumption: non-representativeness of such preference data and so-called "preference pollution," which denotes an unintended relationship between system recommendations and the user's post-consumption preference ratings. We provide an overview of these issues and their importance for the design and application of next-generation recommendation systems, including directions for future research.

show abstract

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

Cited by 99 publications

References 123 publications

Sequential Recommendation via Stochastic Self-Attention

Sequential Recommendation via Stochastic Self-Attention

Unintended Bias in Language Model-driven Conversational Recommendation

Recommender systems, ground truth, and preference pollution

Contact Info

Product

Resources

About