PEAR: Pairwise Evaluation for Automatic Relative Scoring in Machine Translation

TL;DR

PEAR introduces a pairwise supervised evaluation metric for machine translation that predicts quality differences between candidate translations, outperforming existing metrics with fewer parameters and benefiting MBR decoding.

Contribution

The paper proposes a novel pairwise evaluation framework for MT quality estimation that improves accuracy and efficiency over traditional single-candidate metrics.

Findings

PEAR outperforms baseline QE models on WMT24 benchmark.

PEAR uses fewer parameters yet surpasses larger models.

PEAR effectively reduces scoring costs in MBR decoding.

Abstract

We present PEAR (Pairwise Evaluation for Automatic Relative Scoring), a supervised Quality Estimation (QE) metric family that reframes reference-free Machine Translation (MT) evaluation as a graded pairwise comparison. Given a source segment and two candidate translations, PEAR predicts the direction and magnitude of their quality difference. The metrics are trained using pairwise supervision derived from differences in human judgments, with an additional regularization term that encourages sign inversion under candidate order reversal. On the WMT24 meta-evaluation benchmark, PEAR outperforms strictly matched single-candidate QE baselines trained with the same data and backbones, isolating the benefit of the proposed pairwise formulation. Despite using substantially fewer parameters than recent large metrics, PEAR surpasses far larger QE models and reference-based metrics. Our analysis further indicates that PEAR yields a less redundant evaluation signal relative to other top metrics. Finally, we show that PEAR is an effective utility function for Minimum Bayes Risk (MBR) decoding, reducing pairwise scoring cost at negligible impact.