Strengthening interactive security using AI, game theory, and NLP to align people in addressing security challenges in southeastern Iran

Sustainable security in southeastern Iran, especially Sistan and Baluchestan province, requires coordination between local people and security institutions against internal threats such as armed groups, fuel and goods smuggling, theft and banditry, climate challenges, and religious differences. In this study, a framework based on artificial intelligence (AI) and game theory is presented that predicts, optimizes, and aligns the behavior of different actors using cooperative and non-cooperative models. The data for this study are collected from the European Union Cybersecurity Reports (۲۰۲۳), the Global Peace Index (GPI), official statistics from the Statistical Center of Iran (۲۰۲۳), and local reports such as the reports of the Iranian Ministry of Interior on armed conflicts and smuggling in Sistan and Baluchestan. The proposed model, using Nash equilibrium, repeated games, and machine learning algorithms such as reinforcement learning (RL), designs mechanisms to create positive incentives and effective deterrence. To increase realism, the model is extended with a text processing layer based on large language models and Stackelberg language games. Simulation results show that the proposed model increases the level of public participation in confronting local threats by ۳۲% and reduces the likelihood of cooperation with threats such as smuggling and armed groups by ۲۷%.

کلیدواژه ها:

Natural language processing ، language games ، game theory ، dynamic deterrence ، reinforcement learning ، large language models