Prediction Of Information Dissemination Relationship Based On Student Behavior Data

Information dissemination matters,both on an individual and group level,and people’s emotions and behaviors can be spread through information dissemination.For college students who are physically and mentally immature,they are more sensitive and susceptible to unnormal information like rumors.Friendship relationship between students is an indispensable part in the process of information dissemination.Combining friendship network to explore the patterns of information dissemination among students and predict the relationship of information dissemination can identify some students with abnormal behavior and withdrawn personality,and can help college administrators to provide better personalized guidance to students.However,current researches focus on large-scale online message sharing networks like Facebook and Twitter,rather than profile the information dissemination on campus,which fail to provide any references for daily campus management.Against this background,this paper first explores the patterns of information dissemination among students from the facial and psychological perspectives,combining with friendship network,and mainly analyzes the patterns of good/bad information dissemination respectively by using the relevant theoretical knowledge of network science and statistics.Based on some analysis conclusions,the second work of this paper did other extended experiments to predict the relationship of campus information dissemination,which made the framework suitable for large-scale college students,and proposed a framework of generating campus information dissemination network based on multimodal data,called ANSWER(c Ampus i Nformation di Ssemination net Work g Ene Ration).The construction of the ANSWER is listed as four steps.First,we use a convolutional autoencoder to extract the students’ facial features.Second,we process the behavior data to construct the friendship network.Third,heterogeneous information is embedded in the lowdimensional vector space by using network representation learning to obtain embedding vectors.Fourth,we use the deep learning model to predict.The model is trained with two different labels,good news dissemination relation and bad news dissemination relation,and the prediction framework of good news dissemination relationship ANSWER1 and bad news dissemination relationship ANSWER2 are obtained.The results of extensive experimental analysis show that friendship network and good/bad information dissemination network are inextricably linked,similar but inconsistent;Moreover,other’s psychological and facial features can obviously influence a person’s judgment,thus affecting the candidates for information dissemination.Experimental results also show that ANSWER outperforms other methods in multiple feature fusion and prediction of information dissemination relationship performance.