Event-Extraction

Why we included this project

This repository traces an entire Chinese legal-text pipeline rather than a single model. It starts with traffic-accident judgment documents and works through segmentation, part-of-speech tagging, named entity recognition, and CRF-based event-argument extraction, then on to outcome prediction and similar-case retrieval. The directories follow the processing steps instead of a standard package layout, which makes it a good teaching example of how classical NLP tools like LTP, CRF++, jieba, and word2vec fit together before deep learning takes over. Researchers and students in legal informatics or Chinese event extraction can study each stage on its own, thanks to concrete scripts and cached intermediate results. Teams building their own judgment-document analytics should treat it as a reference architecture, not a drop-in service: it targets one case type and needs external tooling like pyltp and CRF++.