Robust Deep Reinforcement Learning for Extractive Legal Summarization

TL;DR

This paper introduces a reinforcement learning approach with novel reward functions to enhance deep summarization models for legal texts, achieving significant improvements across multiple datasets.

Contribution

It presents a reinforcement learning framework with new reward functions tailored for legal summarization, improving existing deep models' performance.

Findings

Significant performance gains on 3 legal datasets

Reinforcement learning outperforms traditional training methods

Novel reward functions effectively balance lexical and semantic quality

Abstract

Automatic summarization of legal texts is an important and still a challenging task since legal documents are often long and complicated with unusual structures and styles. Recent advances of deep models trained end-to-end with differentiable losses can well-summarize natural text, yet when applied to legal domain, they show limited results. In this paper, we propose to use reinforcement learning to train current deep summarization models to improve their performance on the legal domain. To this end, we adopt proximal policy optimization methods and introduce novel reward functions that encourage the generation of candidate summaries satisfying both lexical and semantic criteria. We apply our method to training different summarization backbones and observe a consistent and significant performance gain across 3 public legal datasets.