Hierarchical Attention-based Fuzzy Neural Network for Subject Classification of Power Customer Service Work Orders

Subject classification of power customer service work orders is an important part of power system customer service. However, at present, the power industry is still at the stage of relying on manual classification. Therefore, it is particularly important to propose a method for fast and accurate automatic classification of power customer service orders. In this paper, a Hierarchical Attention-based Fuzzy Neural Network (HABFNN) model based on is proposed for the subject classification of power customer service work orders. The model consists of two parts: the word-sentence layer, which uses the Convolutional Neural Network (CNN) trained by Dense-Fuzzy-Dense (DFD) to model sentences on the basis of word vector representation; the sentence-document layer, which uses CNN and Bidirectional Long Short-Term Memory (BiLSTM) network with attention mechanism to model documents on the basis of the sentence representations. The model can extract important information from the text of power customer service work orders, mine the semantics of the text and classify it. The model is compared with existing text classification methods through experiments. The experimental results show that this model has higher classification accuracy and F1 score, can better classify power customer service work orders and can effectively improve the efficiency and quality of power industry customer service.