Research on Text Classification and Data Analysis of Complaints by Shanghai Gas Company

Since 2018, Shanghai Gas Company has entered customers' complaints by 12345 hotline into the company's platform system. This article uses customers' complaints from 2018 to 2020 for data analysis. Thee complaint text classification based on the improved naive Bayes algorithm to solve the three major points: in the process of manually classifying the complaint text, there are defects and errors caused by human negligence, this is accuracy problem. Due to the large amount of complaint data, manual classification of text data requires a lot of time and effort, this is efficiency problem. The text classification and analysis of the complaint text by human is not deep enough to accurately and quickly locate the hot issues in user complaints, this is depth problem. In view of the above three problems, an improved naive Bayesian algorithm is proposed, this algorithm combines n-gram algorithm and naive Bayes algorithm to realize the classification of complaint texts and solves the disadvantage of naive Bayes not considering word order.