Identification And Prewarning Of Cyber Group Events Based On Average Influence

In recent years,the frequent occurrence of cyber group events has brought great threats to the cyberspace order.The identification and prewarning of cyber group events help to standardize cyberspace management and maintain cyberspace security.Taking Sina Weibo as an example,this thesis analyzes the main problems existing in the current cyber group events identification and prewarning research,and conducts research work on cyber group events identification model and early warning method.The main contents include:Firstly,for the lack of technical definition in current cyber group events research,combined with the existing definition of social science,and analyzing the unique characteristics of cyber group events,the definition of cyber group events is proposed,and the cyber group event recognition model is built.In the feature extraction part of the model,the existing identification method is incomplete to describe the characteristics of the cyber group event.Based on some common features,the characteristics of average influence,verified user ratio and paying user ratio are proposed.In order to find an algorithm more suitable for cyber group event recognition,logistic regression,support vector machine and decision tree algorithm are selected as candidate recognition algorithms.The experimental results show that the feature set selected in this thesis can express the influential features and group features more effectively.Logistic regression algorithm is more suitable for cyber group event recognition in the current data size.Secondly,aiming at the difficulty of quantifying some indicators in the prewarning index system of cyber group events,a set of prewarning index system and classification method of cyber group events are proposed.First,the hidden attributes in the data set are mined,and the prewarning index system is established under the principle of measurability and minimum completeness.Then,according to the probability value of the identification model output,a classification method is developed to classify the cyber group events.Finally,three sets of prewarning index system and two typical cases of cyber group events are selected to conduct comparative experiments by drawing prewarning evolution curve and event trend map.The experimental results show that the evolution curve of prewarning basically conforms to the event trend map,and the prewarning can be accurately carried out at critical time point.