A method for power supply service evaluation based on multimodal public opinion analysis

Abstract

A power supply service evaluation method based on multimodal public opinion analysis relates to a power supply service evaluation method. At present, the content of customer appeals is mainly text data, which has a large amount of information and is unstructured. The invention comprises the following steps: performing voice feature emotion recognition based on the increasing and decreasing component method and the method of long short-term memory network; and performing text appeal classification based on word2vec and LSTM. In this technical solution, the feature extraction of voice and the combination of text features can effectively improve the accuracy of the model's emotion recognition; after word segmentation and feature extraction of the text, the model can better mine the appeal of the text by combining the feature information of the voice Useful information, use the deep recurrent neural network with attention mechanism to pay attention to different keywords at different times, so that the model can better capture the semantic relationship between contexts, mine the timing information and semantic information in the target text, thereby improving text classification the accuracy rate.

Description

technical field [0001] The invention relates to a power supply service evaluation method, in particular to a power supply service evaluation method based on multimodal public opinion analysis. Background technique [0002] Power supply service is a service provided by power companies in order to meet various related needs of customers in obtaining and using power products during the power supply process. Under the background that the quality requirements of power supply service of electric power enterprises are constantly improving with the advancement of power system reform, it is imperative to improve the evaluation method of power supply service. The establishment of a more effective and accurate analysis model based on user feedback is a good starting point for improving the evaluation method of power supply services. Because the text feedback and voice feedback submitted by users are important channels for companies to understand customers, mine customer demands and im...

AI Technical Summary

Problems solved by technology

For the text classification task requested by customers, decision trees, KNN (K-Nearest Neighbor) algorithms, support vector machines, etc. are usually used, but decision trees have problems such as ignoring the correlation between attributes in the data set and overfitting; Interpretability is not strong; support vector machines are sensitive to missing data, there is no general solution to nonlinear problems, etc.

Images