Bradley-Terry Rankings for Recommender Systems Across Dataset Taxonomies
For practitioners and researchers in recommender systems, it provides a more reliable and data-driven approach to algorithm comparison, addressing the limitations of naive metric aggregation.
The paper addresses the challenge of ranking recommendation algorithms fairly across diverse datasets, proposing a Bradley-Terry-based methodology that yields robust rankings dependent on dataset statistics and enables ranking on unseen datasets without model execution.
The ranking of recommendation algorithms is a challenging problem since model performance is sensitive to dataset characteristics such as sparsity, sequential structure, and scale. This drives a demand for a proper methodology for fair comparison between algorithms. Naive aggregation of performance metrics (e.g., averaging NDCG over benchmarks) can yield misleading rankings, undermining practical selection. To address this problem, we introduce a novel, data-driven ranking methodology based on Bradley-Terry (BT) model. We demonstrate that the obtained ranking depends on key dataset statistics. Additionally, we propose a novel metric for evaluating ranking consistency and demonstrate robustness of our ranking to incomplete data. Finally, we introduce a dataset-specific methodology for ranking algorithms on unseen datasets without running the models, relying on extensions of the Bradley-Terry framework, including BT trees and BT models with covariates.