lexpredict-lexnlp

Why we included this project

LexNLP is a Python library built for the kind of legal text that actually shows up in the wild: contracts, plans, policies, and similar material. Where generic NLP tools tend to stumble, it ships a sentence parser that knows legal abbreviations like LLC. or F.3d, along with pre-trained models for splitting documents into pages and sections. The extraction layer pulls out monetary and non-monetary amounts, percentages, dates and recurring dates, durations, and references to courts, regulations, and citations, which covers a lot of what contract review and due diligence actually need. It also includes pre-trained classifiers for document and clause types, plus word-embedding and topic models tuned for specific practice areas, so teams can get useful signals without training everything from scratch. The hundreds of unit tests drawn from real documents are a handy reference for what the library should do on messy input.