NLP Based Latent Semantic Analysis for Legal Text Summarization

It is very essential for lawyers and ordinary citizens to do an exhaustive research related to their case before they answer questions in court. For quite some time they have had to read extremely long judgements and try to pick out the useful information from them or hire legal editors to create summaries. We propose an automated text summarization system that generates short and useful summaries from lengthy judgements. We make use of a natural language processing technique called latent semantic analysis (LSA) to capture concepts within a single document. We use two approaches-a single document untrained approach and a multi-document trained approach depending on the type of input case (criminal or civil). Our data was collected from official government sites that included Supreme Court, high court and district court cases and our model achieved an average ROGUE-1 score of 0.58. Finally, our system was approved by professional lawyers. In the future we aim to provide better continuity within our generated summaries and evaluate our system more accurately.