<![CDATA[This study proposes a Dynamic Latent State Modeling (DLSM) framework to predict public behavior within rapidly evolving digital ecosystems. As online interactions grow increasingly complex shaped by algorithmic exposure, platform norms, and sociopolitical events traditional static models fail to capture the fluidity and nonlinearity of user behavior. Using a combination of Hidden Markov Models, state-space modeling, and probabilistic clustering, this research identifies latent behavioral states underlying observable digital activities such as posting frequency, sentiment shifts, network engagement, and information consumption patterns. Results reveal four major latent states Passive Observation, Selective Engagement, Active Participation, and Reactive Mobilization each corresponding to meaningful psychological and social modes of online behavior. Transition matrices demonstrate that users shift states in response to contextual triggers including emotional content exposure, social reinforcement, platform incentives, and external offline events. The DLSM framework outperforms baseline machine learning classifiers by capturing temporal dependencies and hidden motivational structures influencing online actions. The study offers important implications for digital governance, policy design, crisis communication, marketing strategy, and misinformation management, particularly in anticipating rapid escalations in public sentiment or mobilization. However, limitations include potential dataset biases, constraints on generalizability across platforms, and challenges in detecting synthetic or automated behavior (bots) embedded within user streams. Overall, the research contributes a robust, interpretable, and dynamic approach to understanding and predicting public behavior in complex digital environments.]]>
Copyrights © 2024
