KGCNN: A New CNN with Keywords Group for Crime Classification Over Legal Articles
Chunyang Xing, Liutong Xu, Pengfei Wang · 2018
In juridical field, some keywords in evidence, such as murder, robbery, are quite valuable for judges to classify the evidences. Feeding these keywords in evidences with other words into a united model may weaken the significance of these keywords. In order to overcome this weakness, we propose a simple modification to the Convolutional Neural Networks (CNNs) architecture to better use both of these keywords and other words. In this paper, we present a new convolutional neural networks(CNNs) with keywords group for crime classification over legal articles. Unlike previous methods, our approach emphasizes those important keywords and integrates them into a CNN model. Our models achieve better performance on our legal articles dataset.