Research On The Prediction Of Reposting Behavior Of Microblog Users Based On Planned Behavior Theory

The maturity of mobile Internet technology and the wide application of intelligent terminals drive people into the era of personal Internet.The emergence of the personal Internet era stems from the rise of social media,and the well-known social media platforms are Twitter,Facebook,Weibo,We Chat,etc.With the continuous expansion of user scale and the update speed of weibo data is accelerating,as one of the important carriers for disseminating information,weibo has become a channel for the fermentation of public opinion events.At present,researches on user reposting behavior are analyzed and predicted from three aspects: content characteristics,users,network topology and network characteristics.On the one hand,there is a lack of analysis on the carrier of micro-blog platform;on the other hand,there is a lack of research on user affective preferences.Therefore,in view of the above problems,this thesis establishes the influencing factors model of weibo users' reposting behavior from three different aspects based on the theory of planned behavior,and puts forward a series of research hypotheses based on the analysis of weibo users' behavior and weibo information dissemination mechanism.Firstly,according to the research needs,a certain data acquisition scheme is designed to complete the acquisition of the data set required by the research.Secondly,on the basis of text information pre-processing,the topic distribution of user interest and weibo text are obtained by LDA topic model,and the semantic similarity between them is further calculated.The emotional value of user and weibo is calculated based on emotional dictionary,and the distance between them is utilized to measure the degree of emotional similarity.According to the logon time of weibo users,the number of users concerned and the length of time for weibo users to browse weibo to determine whether users can browse the messages issued by superior users in time.By calculating the specific values of each factor in the influencing factor model and importing the data into SPSS,the hypothesis proposed above is verified and analyzed based on binary logistic regression model.Finally,by comparing the classification results of common classification algorithms,it is found that the sensitivity of logistic regression is the highest.Therefore,logistic regression is used as the basic method to obtain the parameters of the prediction model through training set data and complete the prediction of user reposting behavior.From the research results,it is found that the influencing factors of weibo users' reposting behavior include reposting activity,the semantic similarity between user interest and weibo text,degree of interaction,fans,the emotional similarity between user and weibo.Based on the above factors,the weibo user reposting behavior prediction model is established and the prediction accuracy of the model for reposting behavior reaches 82.65%..The AUC value of model is 0.84,which indicates that the model has good prediction performance.