A Patent Text Classification Model Based on Multivariate Neural Network Fusion
Hongbiao Lu, Xiaobao Liu, Yanchao Yin, Zhicheng Chen · 2019
In order to improve the efficiency and accuracy of automatic classification of patent texts, a patent text classification model (C3-BIGRU-AT) based on multivariate neural network fusion was proposed. Firstly, patent text is segmented and represented by text preprocessing. Then, the text features of different levels and different characteristics are extracted through word embedding layer, convolution layer, BIGRU layer and Attention layer, and text category recognition is carried out through soft Max layer. Finally, case studies show that C3-BIGRU-AT model has a high ability of patent text recognition, and can meet the requirements of accurate and efficient classification of a large number of patent texts.