Reasoning on Security Risks with a Big Data-Based Eventic Graph

With the open communication of the Internet, mass incidents play an increasingly important role in the fermentation of social security risk events. Mass incidents are triggered by social contradictions and are extra-institutional activities with obvious interest demands, based on certain purposes, by specific or unspecified coincident groups with common interests. They are large-scale gatherings, legal or illegal, to express interest demands and policy propositions, and have a certain impact on social order and stability [1]. This report mainly studies the early warning methods for mass incidents such as gatherings and protests. By comparing a variety of efficient machine learning and neural network technologies, combined with the event evolution law mining technology based on the event graph, and using the public negative data of education and training and real estate on the Chinese Internet, a mass incident risk prediction model under incomplete information conditions is proposed. The model generates an abstract event graph of the evolution law of education and training institutions and real estate entities across the country, and outputs factual event graphs for the top ten education and training institutions and the top ten real estate institutions in a certain region. Through model test verification and prediction analysis, this study proves the algorithm effectiveness of the event graph in mass incident prediction.