A customer satisfaction evaluation method of the power company using sentiment analysis

Abstract

This study proposes a method for improving customer satisfaction evaluation of power supply enterprises using sentiment analysis. First, a corpus of comments and posts about power supply enterprise is generated. Second, several machine learning algorithms are used to extract features from the corpus. Finally, the best classifier is used to evaluate the work of a company through the comments and reviews. The experiment shows that Multinomial naive Bayes is more appropriate for clustering the sentiment of Weibo posts. And this approach can benefit the power supply companies to improve their service.