Early detection of unrest signals in social media A case study of Telegram regarding the protests in 1397 SH Iran

Since the advent of the mobile data network, Iran has more than 35 million social media users. Developing social media among the general public (with a penetration rate of 49% in Iran) and their diverse functions can be the starting point for using social sensing. In this case, users provide data, like a sensor, for analysis. One of the applications of social sensing is crisis management. To achieve this goal, big data collection as a defense system can reduce the human, economic and social costs of crises and events, and can be used as a tool to raise situational awareness and enhance national security. The data set of the present study was collected based on the crawling and text mining of verbal violence in one million and four hundred thousand general Persian channels of Telegram messenger in 1397 SH and after refinement, it was modeled based on the time series of the moving average. To identify crisis signals in this model, the oscillator following the momentum trend (which is mostly used in financial analyses) and the moving average of divergence convergence (MACD) are analyzed. This is the first time in the computational social sciences that this tool has been used to predict security crises and political events and to allow government surveillance.The research findings confirm that at least six social protest in the country in 1397 SH were identifiable and manageable before the event. In addition, a system that can use such analyses in social media big data in real-time would have the necessary efficiency to issue an early warning and to measure the political and security risks of society.