Research On Risk Assessment And Early Warning Of University Network Public Opinion Under The Background Of Big Data

With the rapid development of big data technology and mobile Internet,China has entered the network information era of big data,and the Internet is playing an increasingly important role as a platform for information dissemination and opinion expression.University students,as a social group with extremely high network utilization rate and active thinking,are the main force and important influence object for the generation of network public opinion.The frequent occurrence of network public opinion events in colleges and universities has also attracted extensive attention from the public and all walks of life,and the situation of public opinion in universities is becoming increasingly severe.Monitoring the big data information of university network public opinion,and establishing a risk assessment and early warning system are extremely important for quickly and accurately identifying the risk level of university network public opinion,realizing public opinion early warning,and improving the supervision efficiency,coping ability and governance level of universities and related government departments for online public opinion.meaning.Based on the background of big data,we study the risk assessment and early warning of university network public opinion by using big data technology and mathematical model:On the basis of studying the influencing factors,type characteristics and development and evolution law of university network public opinion,the principles of scientific,quantitative,comprehensive and hierarchical are followed.The risk assessment index system of university network public opinion is constructed from four dimensions: reflecting public opinion ontology,public opinion subject,public opinion object,public opinion event force of network environment,netizen force,network media force and network space force.Random Forest algorithm is used to analyse the risk assessment index screening,introducing the Entropy Weight method to determine index weight,the integrated use of TOPSIS method and Grey Relational Analysis method to construct risk assessment model.By weighting the posting progress and grey correlation degree,a crisis risk index reflecting the risk degree of university network public opinion is constructed,and the risk of public opinion is systematically and comprehensively evaluated.Using K-Means algorithm to cluster the crisis risk index,establish and divide the risk level and threshold range of university network public opinion,and finally divide the university public opinion risk into four categories: light alarm,moderate alarm,severe alarm and special alarm,and put forward reasonable and effective management and guidance measures for the university network public opinion with different risk levels.After completing the risk assessment of network public opinion in universities,we will further give early warning to university public opinion.Further to the traditional crossover operator and mutation operator in genetic algorithm adaptive adjustment,the improved Adaptive Genetic Algorithm was applied to the training of the BP neural network and optimization,the AGA-BP neural network model is constructed to predict the crisis risk index,and realize the risk early warning of the university network public opinion.Through the empirical research on the case data of the university network public opinion events in the past four years,the results of the comparison of the early warning model are verified and found that the university network public opinion early warning model based on AGA-BP neural network has extremely high accuracy and precision.Using big data technology,this research greatly simplifies the identification procedures of university network public opinion risk assessment and early warning through scientific methods and accurate and efficient models,and provides a new reference for universities and relevant government departments to deal with,manage and guide university network public opinion reasonably and effectively.