CoPERLex: Content Planning with Event-based Representations for Legal Case Summarization

TL;DR

CoPERLex addresses the challenge of producing faithful, coherent summaries of long legal judgments by decoupling content selection, event-based planning, and summary realization. It combines MemSum-based content selection with an event-centric planning stage that outputs SVO triples to guide a Longformer-based generator, using hybrid training to reduce exposure bias. Across four datasets, the approach yields consistent gains in faithfulness and coherence, especially for long documents, underscoring the value of explicit planning in legal NLP. The work suggests that event-centric representations capture the narrative flow of legal reasoning and offers a pathway for more controllable, accurate legal summarization.

Abstract

Legal professionals often struggle with lengthy judgments and require efficient summarization for quick comprehension. To address this challenge, we investigate the need for structured planning in legal case summarization, particularly through event-centric representations that reflect the narrative nature of legal case documents. We propose our framework, CoPERLex, which operates in three stages: first, it performs content selection to identify crucial information from the judgment; second, the selected content is utilized to generate intermediate plans through event-centric representations modeled as Subject-Verb-Object tuples; and finally, it generates coherent summaries based on both the content and the structured plan. Our experiments on four legal summarization datasets demonstrate the effectiveness of integrating content selection and planning components, highlighting the advantages of event-centric plans over traditional entity-centric approaches in the context of legal judgements.