A BERT and TextCNN integration-based Method for Public Complaints and Proposals Text Classification

With the wide use of Blockchain technology and online public complaints and proposals (hereinafter referred to as PCP) platforms, huge amounts of data are accordingly produced, which poses a challenge on the government's effectively choosing and classifying the key information. In view of this, this paper endeavors to conduct research on the PCP text classification for a smart government. Firstly, PCP texts are collected automatically and preprocessed, and then a hierarchical classification structure involving targeted government departments and PCP contents is proposed. Finally, a BERT and TextCNN model is adopted for PCP text classification. The experimental results show that the classification structure and model proposed in this paper can effectively categorize the PCP text, providing a scientific analysis and effective technological method to improve the work efficiency and service quality of PCP.