Abstract

Summarizing legal case documents is a very daunting task as law practitioners have to traverse via a hundreds of case reports before determining the relevant case which may aid as a good resource material in an ongoing case. Numerous summarization algorithms have been proposed over time, including generic text summarization and a few specifically for summarizing legal documents. This paper proposes different hybrid text summarization techniques by the fusion of extractive and abstractive text summarizer for legal domain. The fusion techniques are aimed at generating high-quality abstractive summaries that balance original content preservation with enhanced readability and coherence. The extractive methods namely LSTM-based summarization, CaseSummarizer, and Maximum Marginal Relevance (MMR), are explored in conjunction with Legal Pegasus, a state-of-the-art abstractive summarization model tailored for legal texts. Evaluation using ROUGE metrics demonstrates the effectiveness of these hybrid approaches in generating coherent and contextually relevant summaries within the legal domain. The CaseSummarizer and Legal Pegasus combination emerges as the most promising, achieving the highest F-score across all evaluated metrics, highlighting its potential for improving legal text comprehension and summarization. The proposed approach aims to elevate performance and deliver summaries that are not only contextually relevant but also more comprehensive.