| With the acceleration of our cultural construction,the demand for knowledge culture is increasingly strengthened,and more and more users rush into the public library to enjoy the edification of book aroma.Diversified social media provides great convenience for people to release information.Different users will have different opinions when experiencing library services,and actively share these opinions on the Internet,which lays a solid data foundation for analyzing the service quality of public libraries.Therefore,exploring the online word-of-mouth of our public library can deeply understand the user demands,provide suggestions for the future construction of public library,and be of great significance for the development of public library.The existing research on library online word-of-mouth mostly uses data from Weibo,post bar,forum and other platforms.The data noise is large,and the accuracy of the evaluation is difficult to guarantee.Based on users’ comments on Dianping,this study will use emotion analysis method to analyze users’ feelings about the service content of public libraries,analyze the pros and cons of the service quality of public libraries,and put forward targeted improvement strategies to improve the service level and quality of public libraries.The research includes the following four parts:(1)The evaluation of Dianping website was selected as the data source,and 13 public libraries were selected as the research objects.Python crawler is used to obtain all review data before October 31,2022 and basic library information including library name,opening hours,official rating of Dianping,total number of reviews,number of thumbs up,number of good and bad reviews,and user ratings(including books,environment and services).After cleaning,preprocessing,word cutting and other processing of the acquired data,word frequency statistics are made to establish the public library online word-of-mouth emotion analysis lexicon.(2)Text classification of online word-of-mouth evaluation is realized based on Word2 Vec model and LDA model.The Chinese corpus of Wikipedia is used to train the Word2 Vec model,and the comments are converted into word vectors,text classification is carried out,words with high similarity are extracted,and the extraction results are input into the LDA model for topic extraction.Six topics are obtained: reader service,collection resources,in-library environment,infrastructure,out-of-library environment,and staff.Based on the results of previous studies,a public library online word-of-mouth evaluation index system corresponding to the threshold value is constructed.According to the word frequency of subject words under each index,the proportion of relevant texts corresponding indicators is calculated,and users’ attention to the service content of public library is analyzed.(3)The SNOWNLP model was used to calculate the emotional mean value of user comments under each index,and the emotional mean value of different public libraries was compared.According to the high score comments and low score comments,the views and opinions of users on public library service were analyzed,so as to explore the advantages and disadvantages of the service quality of different public libraries.(4)Analyze the advantages and disadvantages of the current public library service,point out the shortcomings of the public library service,and put forward the corresponding improvement strategies.Based on the above analysis results,targeted strategies and suggestions are proposed for the improvement of public library online reputation,such as improving the quality of librarians,strengthening the service consciousness of librarians,strengthening the construction of infrastructure,ensuring basic services,strengthening the construction of collection resources,optimizing the results of collection resources,improving the environmental comfort level and strengthening the construction of supporting facilities.This study carries out online word-of-mouth evaluation of public libraries through emotion analysis of user evaluation.Compared with traditional research methods such as questionnaires,the analysis results are more accurate and make up for the shortcomings of previous research content.On the basis of sentiment analysis method based on sentiment dictionary,Word2 vec model and LDA model are introduced for text classification,and good classification effect is achieved,which has certain reference significance for public library online word-of-mouth evaluation research. |
