UGC Quality Prediction Method Based On Persona

With the advent of the big data era,the amount of Internet data has exploded.People can access the Internet anytime,anywhere,using any form of device to collect,use and disseminate network information.The rapid increase in the amount of information on the Internet has not only brought high-quality content to users,but also brought low-quality content that is disorderly and heterogeneous and difficult to use.How to timely and effectively evaluate and identify low-quality UGC(User-Generated Content)and effectively manage and organize high-quality UGC have a great impact on the healthy development of the Internet information environment.At present,most of the UGC quality evaluation and control research is based on the content of UGC itself.User behavior factors have not received due attention,and there is rarely a combination of content and behavior for comprehensive research.This study intends to screen out the abnormal behavior data of low-quality UGC based on the information behavior of social network users,based on the mining of the relationship between user behavior and UGC quality,and construct individual personae for users by using machine learning algorithms.The persona data is trained to obtain a UGC quality pre-judgment model based on the user's persona,so as to effectively predict the future UGC quality of the user.The full text of the study is divided into six chapters.The first chapter and the second chapter are the literature review of the relevant research and theoretical basis of UGC at home and abroad.The third chapter and the fourth chapter are mainly about the principle and construction method of persona.The UGC quality pre-judgment model based on persona is constructed by analyzing and mining the UGC-related behaviors in the social network,identifying the abnormal behaviors of the user,and further screening out the abnormal behaviors that produce low-quality UGC.The fifth chapter is the empirical link.The data is collected by the web crawler and the data is pre-processed to verify the validity and effect of the model.The sixth chapter is a summary,on the basis of summarizing the full text,the future is expected.Since there is no uniform evaluation standard for the determination of low-quality UGC,there is a certain degree of subjectivity in this study.At the same time,the user's real-time data is not dynamically updated into the persona model,which has a certain degree of influence on the accuracy of the model.Therefore,future research can more accurately identify the abnormal behaviors of users producing low-quality UGC based on the establishment of the unified evaluation criteria of low-quality UGC,and dynamically update the real-time data to the persona pre-judgment model,which can improve the accuracy of quality prediction.