Research On The Influence Of COVID-19 On The Internet Emotion Of Students Based On Text Mining Of Weibo Content

Affected by the COVID-19,current research on emotion usually adopts the form of online questionnaires based on self-reporting methods.Due to the influence of social desirability and participants’ willingness to participate,the validity and sample size of researches on emotions in the form of online questionnaires are all limited to a certain extent.In addition,most of the studies on emotions in the context of the epidemic are cross-sectional studies of emotional states in a specific period,and there are few longitudinal studies on the development and changes of emotions over time.During the epidemic,in order to effectively prevent the further spread of the COVID-19,local governments have implemented home quarantine policies.In this context,the Internet has become an important platform for people to vent their emotions.The Internet is full of a large number of Internet text data resources rich in emotional information,and because of the anonymity and strong timeliness of the Internet,the emotions contained are more authentic.The emergence of big data technology has also provided effective tools for the analysis and processing of Internet texts.Starting with online texts to measure and analyze online emotions can more accurately reveal the emotional state of the student population during the epidemic,and better understand the development and change of emotions over time.Sentiment analysis using text mining is adopted in this research which uses the microblog text of the Sina Weibo platform as the research data to measure and analyze the internet emotion of the student population during the epidemic.Using web crawlers,a total of 71996 relevant effective Weibo texts from February 1st to September 1st were collected.Using the sentiment analysis system Senta provided by Baidu AI open platform to analyze the Weibo text,and outputting the sentiment value and sentiment classification results of each Weibo.Randomly select some microblogs for manual annotation,and evaluate the performance of the sentiment analysis model.The Precision index is0.857,the Recall index is 0.912,the F-1 value is 0.884,and all the indexes show that the model performance is good;the manual annotation results are consistent with the sentiment analysis results and The Cronbach alpha coefficient value is 0.844,which has good reliability.The results of the study on the internet emotion of the student population during the epidemic showed that the overall internet emotion of the student population was positive during the epidemic,and the negative internet emotion was obvious in February and March;the distribution of emotional values was relatively large,and there was a phenomenon of emotional extremes;the results of word frequency statistics shown that the negative emotions of the student population during the epidemic mainly come from the long-term home isolation status,online classes and the epidemic,and the positive emotions mainly come from the hope and expectation of the beginning of school and the resumption of classes.Through the analysis of the daily microblog posting volume,it is found that there are 5 outburst nodes of network sentiment,of which 3 nodes have corresponding triggering events,and the remaining nodes have no triggering events.Combined with the results of the network sentiment value of the node date,it is found that there is a certain spontaneity in the outbreak of network sentiment,that is,without a specific triggering event,the amount of network text published on the day is large,the proportion of positive sentiment network text and the sentiment value are more extreme.Based on the differences in the distribution of emotional values in different stages,the development of student groups’ online emotions during the epidemic is divided into four stages: the initial period,the outbreak period,the recovery period,and the growth period.Except that there is no significant difference in the distribution of emotional values between the initial period and the recovery period(P=0.926),there are significant differences in the distribution of emotional values between other stages(P<0.001).During the outbreak period,the number of Weibo posts was the largest and the average emotional value was the lowest;during the growth period,the average emotional value was the highest.The stages of the emotional development of the student population during the epidemic have provided timely reference and effective detection methods for schools at all levels to provide timely mental health education and psychological interventions to students.