Quantitative Risk Scoring of Loan Documents using Weighted Keyword Matching and Semantic Similarity

LegalSutra is a rule-based system that automatically analyzes loan documents to assess legal risk. It identifies predefined risk phrases and proposes to use semantic similarity to detect variations. The system penalizes missing key clauses, such as repayment or termination terms, which may signal hidden risks. It outputs a normalized risk score—categorized as low, moderate, or high—and includes encryption and summarization features for secure, scalable, and consistent legal risk evaluation without needing large labeled datasets.