Research And Application Of NLP In The Court's Intelligent Trial System

With the speeding up of China's building of a strong legal system,the courts are facing increasing data on the types of cases,persons involved,and judgment results.In the course of the court's trial of the case,in order to avoid major errors such as similar judgments and different judgment results,case adjudicator need to spend a lot of time reading the case transcript and analyzing the related historical cases when making a judgment on the case.This not only causes a waste of manpower and material resources,but also inevitably causes some misjudgments and omissions in the process,which will cause certain losses to the country and people.Based on this kind of problems encountered by the people's court,this thesis uses Natural Language Processing and Deep Learning technology to intelligently generate the corresponding judgment results according to the court,s written description of the criminal case,and convert the judgment results into the court,s standard judgment documents.Court judges use this adjudication document as a reference,so they can quickly and accurately make judgments on cases and give fair and just judgment results.The ruling documents serve as the final basis of the court's trial results,including the description of the case and the judgment results.Therefore,this article first uses crawler technology to obtain nearly 200,000 adjudicative documents that have been published from the website of a referee document of a domestic court.Then it use regular expression and other technologies to extract the key information in these documents to build a corpus in JSON format.And it use the relevant knowledge of Natural Language Processing to perform word segmentation and text vectorization on the corpus,and input it into the neural network based on Deep Learning.After multiple rounds of training,we obtain a prediction model of the eourt,s decision and apply the model to the eourt,s intelligent trial system.The court,s intelligent trial system takes the text description of the case details as an input,which i5 passed into the eourt,s judgment result prediction model,and the model will predict the judgment result of the case.The predicted judgment results include the crimes committed by the suspect,the sentence of the sentence and the relevant legal provisions on which the judgment was made.We finally used these predicted judgment results and 'intelligently generated a judgement document according to the fiamat requirements of the judgement document.In the research process of thio thesis,not only the basic knowledge of natural language processing is used,but also a Deep Learning-based NLP text classification m odel is established,and a more optimized text classification model is proposed.It also makes the accuracy of prediction decision results has been improved,and at the same time,the efficiency of the judgment of relevant court personnel has been improved.