From Bias to Balance: Fairness-Aware Paper Recommendation for Equitable Peer Review
#Fair-PaperRec #Bias mitigation #Academic equity #Machine learning #Double-blind review #Underrepresented groups #Conference selection #Algorithmic fairness
📌 Key Takeaways
- Fair-PaperRec is a machine learning system that addresses systemic biases in academic paper recommendations
- The system achieved up to 42.03% increase in underrepresented-group participation with minimal impact on quality
- The researchers tested both synthetic datasets and real conference data from major computer science venues
- The system uses a fairness regularizer that considers intersectional attributes like race and country
📖 Full Retelling
Researchers Uttamasha Anjally Oyshi and Susan Gauch introduced Fair-PaperRec, a machine learning system designed to address systemic biases in academic paper recommendations, on arXiv on February 25, 2026, in response to persistent disadvantages faced by underrepresented groups in scholarly publishing despite double-blind review practices. The researchers developed Fair-PaperRec as a Multi-Layer Perceptron with a differentiable fairness loss that considers intersectional attributes such as race and country, which re-ranks papers after the initial double-blind review process to increase representation while maintaining scholarly quality. Through their research, the team first tested their hypothesis on synthetic datasets with varying levels of bias—high, moderate, and near-fair—to understand how fairness parameters affect diversity and utility, demonstrating robustness and adaptability under different disparity levels. In their real-world application, applying Fair-PaperRec to conference data from three major computer science venues—ACM SIGCHI, Designing Interactive Systems, and Intelligent User Interfaces—showed that an appropriately tuned configuration achieved up to a 42.03% increase in underrepresented-group participation while maintaining at most a 3.16% change in overall utility compared to historical selection methods. The researchers concluded that fairness regularization can serve both as an equity mechanism and a mild quality regularizer, particularly in environments with high bias, offering a practical framework for developing equitable AI systems that balance representation with quality.
🏷️ Themes
Algorithmic fairness, Academic publishing, Machine learning
📚 Related People & Topics
Underrepresented group
Population subset
An underrepresented group describes a subset of a population that holds a smaller percentage within a significant subgroup than the subset holds in the general population. Specific characteristics of an underrepresented group vary depending on the subgroup being considered.
Machine learning
Study of algorithms that improve automatically through experience
Machine learning (ML) is a field of study in artificial intelligence concerned with the development and study of statistical algorithms that can learn from data and generalize to unseen data, and thus perform tasks without explicit instructions. Within a subdiscipline in machine learning, advances i...
Entity Intersection Graph
No entity connections available yet for this article.
Original Source
--> Computer Science > Machine Learning arXiv:2602.22438 [Submitted on 25 Feb 2026] Title: From Bias to Balance: Fairness-Aware Paper Recommendation for Equitable Peer Review Authors: Uttamasha Anjally Oyshi , Susan Gauch View a PDF of the paper titled From Bias to Balance: Fairness-Aware Paper Recommendation for Equitable Peer Review, by Uttamasha Anjally Oyshi and 1 other authors View PDF HTML Abstract: Despite frequent double-blind review, systemic biases related to author demographics still disadvantage underrepresented groups. We start from a simple hypothesis: if a post-review recommender is trained with an explicit fairness regularizer, it should increase inclusion without degrading quality. To test this, we introduce Fair-PaperRec, a Multi-Layer Perceptron with a differentiable fairness loss over intersectional attributes (e.g., race, country) that re-ranks papers after double-blind review. We first probe the hypothesis on synthetic datasets spanning high, moderate, and near-fair biases. Across multiple randomized runs, these controlled studies map where increasing the fairness weight strengthens macro/micro diversity while keeping utility approximately stable, demonstrating robustness and adaptability under varying disparity levels. We then carry the hypothesis into the original setting, conference data from ACM Special Interest Group on Computer-Human Interaction , Designing Interactive Systems , and Intelligent User Interfaces . In this real-world scenario, an appropriately tuned configuration of Fair-PaperRec achieves up to a 42.03% increase in underrepresented-group participation with at most a 3.16% change in overall utility relative to the historical selection. Taken together, the synthetic-to-original progression shows that fairness regularization can act as both an equity mechanism and a mild quality regularizer, especially in highly biased regimes. By first analyzing the behavior of the fairness parameters under controlled conditions and then valid...
Read full article at source