Rationalizing Text Matching: Learning Sparse Alignments via Optimal Transport

TL;DR

This paper introduces a novel optimal transport-based method for sparse and interpretable text alignment in matching tasks, enabling end-to-end training without explicit alignment labels.

Contribution

It develops constrained optimal transport variants for sparse text alignment, improving interpretability while maintaining prediction accuracy.

Findings

Achieves highly sparse and faithful alignments

Maintains high prediction accuracy with sparse rationales

Outperforms attention-based baselines in interpretability

Abstract

Selecting input features of top relevance has become a popular method for building self-explaining models. In this work, we extend this selective rationalization approach to text matching, where the goal is to jointly select and align text pieces, such as tokens or sentences, as a justification for the downstream prediction. Our approach employs optimal transport (OT) to find a minimal cost alignment between the inputs. However, directly applying OT often produces dense and therefore uninterpretable alignments. To overcome this limitation, we introduce novel constrained variants of the OT problem that result in highly sparse alignments with controllable sparsity. Our model is end-to-end differentiable using the Sinkhorn algorithm for OT and can be trained without any alignment annotations. We evaluate our model on the StackExchange, MultiNews, e-SNLI, and MultiRC datasets. Our model achieves very sparse rationale selections with high fidelity while preserving prediction accuracy compared to strong attention baseline models.