Executive Summary : | Ratio Decidendi is a judicial sentence expressing the essential reasoning or legal principle behind a court's decision. Extraction of ratio decidendi might be difficult, even for legal specialists, in nations with common law systems like India that place more emphasis on precedence. Predictive coding in the legal domain utilizes machine learning techniques for e-discovery to automate the review of electronically stored collections of legal documents. Automating the identification of ratio decidendi in e-judgments is a potential problem in Legal Artificial Intelligence. The models are trained using a small sample of manually reviewed documents and then applied to the rest of the data set to extract the ratio decidendi from precedence. Another challenge is insufficient labelled data for training the machine learning models. In such scenarios, data augmentation will be beneficial to artificially increase the amount of data for training. The idea of data augmentation strategies with a legal expert in the middle can be helpful in generating useful training data. The task of extracting sentences from a single legal document was also proposed. Finally, we compare the model's performance with the state-of-the-art chatGPT model. Predictive coding is intended to reduce the time and cost associated with the manual review of judgements in the extraction of ratio decidendi in Indian judicial decisions. |