Connecting User and Item Perspectives in Popularity Debiasing for Collaborative Recommendation

@article{Boratto2020ConnectingUA,
  title={Connecting User and Item Perspectives in Popularity Debiasing for Collaborative Recommendation},
  author={Ludovico Boratto and Gianni Fenu and Mirko Marras},
  journal={Inf. Process. Manag.},
  year={2020},
  volume={58},
  pages={102387}
}

Novel approaches to measuring the popularity inclination of users for the popularity bias problem

This study proposes two novel approaches, namely Better-Than-Average (BTA) and Positively-Rated (PR), to measure the popularity inclination of individuals more correctly, which aim to consider only items favored by users rather than all rated ones but differ in how they select the set of liked items.

Incorporating Bias-aware Margins into Contrastive Loss for Collaborative Filtering

Considering the theoretical guarantee and empirical success of BC loss, this work advocates using it not just as a debiasing strategy, but also as a standard loss in recommender models.

Exploring potential biases towards blockbuster items in ranking-based recommendations

  • Emre Yalcin
  • Computer Science
    Data Mining and Knowledge Discovery
  • 2022
This study presents a practical formulation that measures the degree of the blockbuster level of the items by combining their liking-degree and popularity effectively, and performs a comprehensive set of experiments to explore the potential biases towards blockbuster items in recommendations.

Countering Popularity Bias by Regularizing Score Differences

This study proposes a novel method to reduce the model bias while maintaining accuracy by directly regularizing the recommendation scores to be equal across items a user preferred by extending the widely used pairwise loss which maximizes the score differences between preferred and unpreferred items.

A Formal Analysis of Recommendation Quality of Adversarially-trained Recommenders

This work model the learning characteristics of the BPR and APR optimization frameworks to give mathematical evidence that, when the feedback data have a tailed distribution, APR amplifies the popularity bias more than BPR due to an unbalanced number of received positive updates from short-head items.

Understanding the Effects of Adversarial Personalized Ranking Optimization Method on Recommendation Quality

This work model the learning characteristics of the BPR and APR optimization frameworks to give mathematical evidence that, when the feedback data have a tailed distribution, APR amplifies the popularity bias more than BPR due to an unbalanced number of received positive updates from short-head items.

The Interaction Graph Auto-encoder Network Based on Topology-aware for Transferable Recommendation

This work proposes an Interaction Graph Auto-encoder Network (IGA) based on topology-aware to address the transferable recommendation problem and demonstrates that IGA outperforms state-of-the-art approaches in transferable scenarios.

Self-Supervised learning for Conversational Recommendation

References

SHOWING 1-10 OF 44 REFERENCES

Managing Popularity Bias in Recommender Systems with Personalized Re-ranking

This paper introduces a personalized diversification re-ranking approach to increase the representation of less popular items in recommendations while maintaining acceptable recommendation accuracy.

Novel Recommendation Based on Personal Popularity Tendency

An efficient novel-recommendation method called Personal Popularity Tendency Matching (PPTM) is proposed which recommends novel items by considering an individual's Personal Popularities Tendency (or PPT), which helps to diversify recommendations by reasonably penalizing popular items while improving the recommendation accuracy.

Popularity-Aware Item Weighting for Long-Tail Recommendation.

This paper examines an item weighting approach to improve long-tail recommendation and finds it works as a simple yet powerful add-on to existing recommendation algorithms for making a tunable trade-off between accuracy and long- tail coverage.

Improving pairwise learning for item recommendation from implicit feedback

The experiments indicate that the proposed adaptive sampler improves the state-of-the art learning algorithm largely in convergence without negative effects on prediction quality or iteration runtime.

Evaluating the Relative Performance of Collaborative Filtering Recommender Systems

An evaluation framework based on a set of accuracy and beyond accuracy metrics, including a novel metric that captures the uniqueness of a recommendation list is presented, which finds that the matrix factorisation approach leads to more accurate and diverse recommendations, while being less biased toward popularity.

Missing Data Modeling with User Activity and Item Popularity in Recommendation

This paper proposes a novel approach to weight the missing data based on user activity and item popularity, which is more effective and flexible than the uniform-weight assumption and outperforms 3 state-of-the-art models including BPR, WMF, and ExpoMF.

Should I Follow the Crowd?: A Probabilistic Analysis of the Effectiveness of Popularity in Recommender Systems

A crowdsourced dataset devoid of the usual biases displayed by common publicly available data is built, in which contradictions between the accuracy that would be measured in a common biased offline experimental setting, and the actual accuracy that can be measured with unbiased observations are illustrated.

What recommenders recommend: an analysis of recommendation biases and possible countermeasures

It is shown that popular recommendation techniques—despite often being similar when compared with the help of accuracy measures—can be quite different with respect to which items they recommend.

Diversity, Serendipity, Novelty, and Coverage

A survey of the most discussed beyond-accuracy objectives in recommender systems research: diversity, serendipity, novelty, and coverage is presented and the positive influence of novelty on recommendation coverage is demonstrated.

Controlling Popularity Bias in Learning-to-Rank Recommendation

This paper introduces a flexible regularization-based framework to enhance the long-tail coverage of recommendation lists in a learning-to-rank algorithm and shows that regularization provides a tunable mechanism for controlling the trade-off between accuracy and coverage.