Analysis And Prediction Of The Spread Of Social Security Incidents Oriented To Social Media

With the development of science and technology,the main channel of information dissemination has gradually shifted from traditional media such as newspapers and television to online media,which makes it more efficient and convenient for people to acquire,publish and transmit information.The online social media represented by micro-blog has created a convenient environment for information flow and data sharing,but it has also accelerated the spread of social security incidents and the online public opinion,and provided a loophole for foreign hostile forces and lawsbreakers to spread rumors and trigger negative public sentiment.So analyzing the social security incidents dissemination and predicting the trend of them in online social media not only play an important role in network opinion monitoring and have important significance for ensuring network space security and maintaining social stability.This paper takes Sina Weibo as the research object,explores and studies the influence evaluation method of social security incidents in micro-blog,the forwarding behavior prediction of users and the propagation paths prediction of incidents.The research work has formed a set of models and systems for the monitoring of social security incidents from the analysis and prediction of social security incidents.The main research contents and results of this paper are as follows:(1)Research on the construction method of micro-blog resource base for social security incidentsThis paper proposes a micro-blog data crawling method for social security incidents.According to the research of web crawler and Sina Weibo platform,the Scrapy crawler framework and Sina Weibo API are used to build a micro-blog data crawling system based on social security incident retrieval conditions.Aiming at the characteristics of network text,a micro-blog text processing strategy is constructed based on the user dictionary and advertisement dictionary to filter the spam micro-blog and segment the text.And finally,use the processed data to construct the micro-blog resource base for social security incidents.(2)Research on the influence evaluation method of social security incident in micro-blogThis paper proposes an influence evaluation method for social security incidents in micro-blog.By analyzing the characteristics of micro-blog information dissemination,a directed graph of social security incident propagation is constructed.Based on the incident propagation graph,a major disseminators’ influence calculation model and an incident propagation amount calculation method are constructed.After that,the method of evaluating the social security incident influence is constructed based on the three aspects of the major disseminators’ influence,the propagation amount and the propagation speed.Experiments show that the major disseminators of this model mining have strong representation,and the incident influence assessment performance is better.(3)Research on the forwarding behavior prediction method of micro-blog userThis paper proposes a set of micro-blog user forwarding prediction feature system and forwarding behavior prediction model in social security incidents.By analyzing the personal interests and behavior habits of users,the forwarding prediction feature system is constructed from the aspects of user characteristics,social characteristics and incident characteristics.Use TF-IDF to extract key words for representing user interests and incident information,and use the word vector and cosine similarity formula to calculate the similarity between users and incidents.Based on the feature system,the SVM classification model is used to construct the forwarding behavior prediction model in social security incidents.Experiments show that the forward prediction feature system characterizes the user’s behavior habits suitably and the model has a higher accuracy in forwarding behavior prediction of micro-blog user.(4)Research on the propagation path prediction method of social security incident in micro-blogThis paper proposes a propagation path prediction model of social security incidents in micro-blog.The social network relationship diagram is constructed through the user’s “follow” relationship in Sina Weibo,and the social security incident propagation process from the initial to the end is analyzed based on the diagram.An IC-PPM propagation path prediction model is constructed based on the independent cascade model propagation analysis,the micro-blog user forward behavior prediction model is improved and the LT-PPM propagation path prediction model is constructed according to the linear threshold model propagation analysis.Experiments show that the linear threshold model can better simulate the true propagation characteristics of social security incidents than the independent cascade model.LT-PPM has higher accuracy than IC-PPM.Finally,the social security incident propagation analysis system is built using the Bootstrap framework and the Vue framework under the SSM architecture.The system consists of five modules: social security incident information database display,incident influence calculation,disseminators analysis,the major disseminators display and propagation path analysis,which can analyze and visualize the dissemination of social security incidents in Sina Weibo.