基于深度学习的法院信息文本分类

Research on Classification of Court Information Texts Based on Deep Learning

为解决在法院数据信息化过程中,海量的法院文书存在缺乏自动管理分类的问题,提出一种基于字符级卷积神经网络的文本分类模型。模型通过卷积神经网络进行特征提取,能够精确有效地解决文本分类问题。实验结果证明,该模型可以实现在测试集上准确率99.67%的分类,且训练用时只有常用循环神经网络算法的50%。

In the process of court data informatization,there is a lack of automatic management classification in massive court documents.This paper proposes a text classification model based on character-level convolutional neural network,which can effectively solve the problem.The model extracts features through convolutional neural networks,which can classify texts efficiently and accurately.Experiments show that the model can achieve an accuracy rate 99.67%of classification on the test set,and the training time is only 50%of the commonly used Recurrent Neural Networks.