پويا صحفي

While most protests remain peaceful, some turn violent an‎d lead to potentially dangerous situations. Given the positive correlation between social media use an‎d protest participation, the aim is to predict the potential occurrence of protests using social media data. In this study, a dataset of tweets related to protests from 2019 to 2023 was collected. Additionally, machine learning models were trained to distinguish tweets related to protest events from those not related to protests. Finally, an early warning system was created that automatically monitors tweets an‎d predicts the potential occurrence of a protest. In this project, various supervised learning models an‎d convolutional neural networks were used, among which the convolutional neural network model performed best with an accuracy of 94.5% in distinguishing tweets related to protest events from unrelated ones. To the extent possible, the developed system using Twitter data was able to detect the occurrence of events during protest demonstrations, but it can be improved.

پيش بيني , تظاهرات اعتراضي , نا‌امني مدني , توييتر , يادگيري ماشين , تحليل شبكه هاي اجتماعي

prediction , Protest Demonstration , Civil Unrest , Twitter , Machine Learning , Social Network Analysis

Pouya Sohofi

Hasan Naderi