| In recent years,with the continuous advancement of artificial intelligence technology,various fields are developing towards intelligence under its guidance.In the judicial field,with the release of a large number of judgment documents,the prediction of legal judgments using natural language processing technology to mine and analyze judgment documents has gradually become a popular concern.Legal judgment prediction aims to predict the verdict of legal cases based on factual description.It mainly includes charge prediction,article prediction,sentence prediction,fine prediction and other aspects.It can not only provide low-cost but high-quality legal consulting services for nonprofessionals,but also serve as a convenient reference for professionals,improve their work efficiency,and alleviate the imbalance between the increase in criminal cases and the lack of grassroots judges.In view of the fact that the current legal judgment prediction model cannot make full use of the information in the criminal fact text,this thesis researches and implements a new deep learning model,which uses ternary joint extraction,heterogeneous graph neural network and multi-task learning to improve the accuracy of legal judgment prediction.At the same time,in view of the fact that there is currently no intelligent system for predicting legal judgments on judgment documents,this thesis designs and implements a legal judgment prediction system for judgment documents,which can predict various legal judgment results for judgment documents of a given charge.At the same time,it can also deal with the situation that multiple defendants and one defendant are accused of multiple crimes in one judgment document.As a system for assisting professionals and nonprofessionals,the system has a wide range of application scenarios.It can not only predict more accurate legal judgments for judgment documents uploaded by ordinary users,but also help administrators perform visual analysis on all predicted judgment documents data.This system uses the Django framework to develop the front-end of the web page and build the Web service,and completes the development of functional modules such as user information management,judgment document management,legal judgment prediction and visual management. |