Applying NLLP and ML to Predict Damages as a Remedy for Contract Breach

Motivated by the subjective decision making and lack of strict protocols in damages as a remedy for contract breach, this project uses natural legal language processing (NLLP) and artificial intelligence (AI) techniques to analyze patterns in contract law cases and reduce uncertainty in their outcome. A 'hybrid' model combining heuristics, NLLP & the results of an LSTM based model into an XGBoost regressor along with contextual information had the best performance for the classification of entity types from unstructured proceedings text. Linear regressors were developed to approximate the Recovery Rate and the Win Rate using a set of 6 engineered features likely to affect the outcome.