Event assigning based on hierarchical features and enhanced association for Chinese mayor's hotline

Nowadays, manual event assignment for Chinese mayor's hotline is still a problem of low efficiency. In this paper, we propose a computer-aided event assignment method based on hierarchical features and enhanced association. First, hierarchical features of hotline events are extracted to obtain event encoding vectors. Second, the fine-tuned RoBERTa2RoBERTa model is used to encode the “sanding” responsibility texts of Chinese local departments. Third, an association enhanced attention (AEA) mechanism is proposed to capture the correlation information of the “event-sanding” splicing vectors for the sake of obtaining matching results of “event-sanding,” and the matching results are input into the classifier. Finally, the assignment department for is obtained by a department selection module. Experimental results show that our method can achieve better performance compared with several baseline methods on HEAD (a dataset we construct independently). The ablation experiments also demonstrate the validity of each key module in our method.