Research On The Impact Of Online Events Based On Social Network Data

As the Internet enters the web 2.0 era dominated by user-generated content,information resources on the Internet are more abundant,the dissemination and diffusion of information is faster and more convenient,and people’s reliance on the use of the Internet is also increasing.Social media platforms such as Weibo,Twitter,etc.provide a large amount of data and information for studying these social network events,including text,audio and video,and record the spread of these events on the Internet.The general method of analyzing the impact mainly considers the number of likes.Natural features such as the number of comments,the number of comments,and the number of reposts,together with some questionnaire surveys,are relatively subjective and cannot truly portray the influence of cyber incidents.This article is supported by the Ministry of Science and Technology’s “Key Research Project on Cyberspace Security” to further develop cyber incidents.Impact quantitative research work.This thesis extracts deep features from the text data of the network event and the communication network,each with emotional features,which represents the public’s emotional tendency towards the event in the network event.The category feature indicates which category the event belongs to,such as government affairs,sports,etc.The subject’s influence characteristic indicates the influence of the publisher of the network event and the influence of the core person in the event.The information entropy feature indicates the amount of information contained in the main post text of the network event.The calculation of each feature uses a corresponding model or algorithm.Through the improved BERT model,a network event sentiment classification and category classification model is constructed.The effect of the model is mainly improved by adding a step of unsupervised learning of a specific corpus before the classification task,and finally achieved 93.9% and 93.0% accuracy respectively Rate,and then calculate the emotional characteristics and category characteristics of the network event based on the model.In order to calculate the main body influence characteristics of the network event,firstly,the main body of the network event is extracted from the text of the event using the technology of named entity recognition,and then the main body influence characteristic is calculated based on the Page Rank algorithm.The calculation of the information entropy characteristics of network events is based on Shannon’s information theory,and the network events are regarded as multi-dimensional random variables to calculate the amount of information of this random variable.After extracting the deep features of the network event,combined with the rest of the natural features such as the number of likes,the number of reposts,etc.,based on the GBDT machine learning model,a supervised learning method is used to implement a quantitative model of the impact of network events.The three indicators of mean square error,average absolute percentage error and coefficient of determination are used to evaluate the performance of the model.The mean square error is equal to 58.981,the average absolute percentage error is equal to 0.033,and the coefficient of determination is equal to 0.954,all of which are better than comparison.Linear regression model.At the same time,the model is used to calculate the influence of cyber incidents in January 2021,and the obtained influence ranking results are compared with the ranking results of People’s Daily Online Public Opinion Data Center and Sina Hotspot.The results obtained show the quantitative model of cyber incidents proposed in this paper.It can describe the influence of network incidents more realistically.