| The establishment of online complaint and suggestion channels by government agencies has greatly expanded and facilitated the public’s need to express their demands to the government.With the increasing number of users of online government services in China,the public’s demand for the quality,variety and speed of services provided by government departments has gradually increased.How to handle people’s complaints and suggestions to government departments quickly and efficiently,improve the quality of government services to the public,enhance the credibility of the government and reduce the level of dissatisfaction of the public has become a matter of great concern to government managers and scholars.In this paper,we use text mining,topic modelling,k-means clustering and other methods to analyse the text of e-government complaints and suggestions.The research and analysis are carried out on the data of complaints combined with the content of complaints and other aspects,which provide reference values for local governments to effectively deal with people’s complaints and suggestions,and improve the quality of urban construction and services of the masses’ surnames.This paper combs the text content of complaints and suggestions in the message boards of Shanxi provincial leaders and Taiyuan government mailboxes from January 2021 to December 2022,with a total of 4723 items,and carries out word pre-processing and deactivation pre-processing on the complaint data through the text visualization method of e-government complaints and suggestions.The word frequencies of the words were calculated,the TF-IDF values of the words were calculated,the paper builds a document word matrix to measure the effect of every word in every document.The k-means clustering algorithm was also used to extract the aspects of people’s complaints that are of more concern to them by clustering the characteristic words of the government complaints suggestions,which provides reference for supporting decision-making.Finally,based on the text of e-government complaints and suggestions,the seven types of complaints that need to be handled as a priority and the intensity of negative emotions of the public in the seven problem types are analysed,which can provide a basis for the government to optimise the design of the complaints and suggestions platform and handle people’s complaints and suggestions by category.In conclusion,the complaint text automatic classifier constructed in this paper,as well as its split-word thesaurus,deactivation thesaurus and sentiment dictionary,are highly innovative for efficiently processing e-government complaint suggestion complaint data,providing tool support and countermeasure suggestions to assist the government in processing e-government complaint suggestions,and,continuously promoting the construction of smart digital government,with good application value. |
