Deconfounding Legal Judgment Prediction for European Court of Human Rights Cases Towards Better Alignment with Experts

TL;DR

This paper introduces a deconfounding approach for legal judgment prediction that leverages expert knowledge and adversarial training to improve alignment with legal experts, reducing reliance on superficial signals.

Contribution

It proposes a novel deconfounding method using adversarial training and provides expert annotations to enhance model validation in legal judgment prediction.

Findings

Deconfounded models align better with expert rationales.

Adversarial training reduces reliance on superficial signals.

Expert annotations improve benchmark evaluation.

Abstract

This work demonstrates that Legal Judgement Prediction systems without expert-informed adjustments can be vulnerable to shallow, distracting surface signals that arise from corpus construction, case distribution, and confounding factors. To mitigate this, we use domain expertise to strategically identify statistically predictive but legally irrelevant information. We adopt adversarial training to prevent the system from relying on it. We evaluate our deconfounded models by employing interpretability techniques and comparing to expert annotations. Quantitative experiments and qualitative analysis show that our deconfounded model consistently aligns better with expert rationales than baselines trained for prediction only. We further contribute a set of reference expert annotations to the validation and testing partitions of an existing benchmark dataset of European Court of Human Rights cases.