| Abstract |
This study introduces the neuroEmotion Classifier, a Sustainable, Smart System-oriented framework for emotion classification in social media text, with applications in urban resilience, crisis management, and community well-being. Designed to analyze tweets related to crises, conflicts, and tragedies, the system maps emotional states to neuroemotion processes, enabling emotion-aware urban analytics for real-time monitoring of public sentiment in high-stress scenarios. By integrating a novel emotion-neuroemotion taxonomy, a specialized lexicon, and semi-automated annotation (Cohen’s kappa = 0.85), the framework bridges linguistic patterns with neurobiological insights to support data-driven urban decision-making in urban planning and public policy. Evaluated across multiple machine learning pipelines, the sustainable emotion detection Smart Urban System leverages LightGBM with Neuroemotion-Informed BERT embedding to achieve 98.57% accuracy, 0.9855 F1 score, and 1.0 AUC, outperforming traditional representations like TF-IDF and Word2Vec. While BERT-based models require greater computational resources, their superior performance justifies their use in urban smart infrastructure for emotion-aware applications, striking a balance between precision and scalability. This work establishes a foundation for deploying Neuroemotion-aware emotion classifiers in smart urban systems, enhancing urban governance, public safety protocols, life cycle aware urban design, and adaptive city planning through contextually rich social media analysis. © 2025 International Journal of Sustainable Building Technology and Urban Development. |
