deep-lawyer is a classifier used on legal judgement prediction task, based on CAIL2018 dataset. we use Multi-perspective Bi-Feedback Network with combines multi-task learning, attention mechanism with traditional textCNN.

Source code and data description of Legal Judgment Prediction via Multi-Perspective Bi-Feedback Network.

All code uploaded contains data_processing and model_training, but does not contain any visualizing tool or baseline code. Probably we will give a specific usage of deep-lawyer code in future update.

Add a brief data processing code description. See /meta/数据描述_服务器.txt.

Legal Judgment Prediction via Multi-Perspective Bi-Feedback Network