Supervised Contrastive Learning Approach for Contextual Ranking

TL;DR

This paper introduces a supervised contrastive learning method with data augmentation to enhance document ranking performance on small datasets, outperforming traditional objectives especially in limited data scenarios.

Contribution

It proposes a novel supervised contrastive loss with data augmentation for ranking, demonstrating significant improvements on small datasets across various domains.

Findings

Supervised contrastive loss improves ranking performance on small datasets.

Data augmentation alone does not necessarily enhance performance with traditional objectives.

The method shows notable gains in news, finance, and scientific fact-checking datasets.

Abstract

Contextual ranking models have delivered impressive performance improvements over classical models in the document ranking task. However, these highly over-parameterized models tend to be data-hungry and require large amounts of data even for fine tuning. This paper proposes a simple yet effective method to improve ranking performance on smaller datasets using supervised contrastive learning for the document ranking problem. We perform data augmentation by creating training data using parts of the relevant documents in the query-document pairs. We then use a supervised contrastive learning objective to learn an effective ranking model from the augmented dataset. Our experiments on subsets of the TREC-DL dataset show that, although data augmentation leads to an increasing the training data sizes, it does not necessarily improve the performance using existing pointwise or pairwise training objectives. However, our proposed supervised contrastive loss objective leads to performance improvements over the standard non-augmented setting showcasing the utility of data augmentation using contrastive losses. Finally, we show the real benefit of using supervised contrastive learning objectives by showing marked improvements in smaller ranking datasets relating to news (Robust04), finance (FiQA), and scientific fact checking (SciFact).