Intellectual Model of Public Appeals Management Based on Digital and Smart Infrastructures

The rapid acceleration of digital transformation and urbanization has made the effective management of public appeals a critical challenge for modern public administration. Most existing electronic appeal systems remain reactive in nature and are designed to respond only after problems have already occurred. This article examines key issues related to digital government, smart city infrastructures, and interactive service architectures, and proposes an intelligent, proactive, and data-driven model for public appeals management.
The proposed model integrates several natural language processing stages, including textual appeal cleaning, language detection, tokenization, and embedding generation. Within this framework, public appeals are interpreted not merely as administrative documents but as socio-technical analytical signals reflecting the real-time state of urban infrastructure and public services.
By combining artificial intelligence methods, geospatial data analysis, and a human-in-the-loop approach, the model enables early problem detection, dynamic prioritization, and flexible decision-making. Complex textual appeals are grouped according to their structural and semantic characteristics, and the most appropriate processing methods are identified for each group. Experimental results demonstrate that hashing-based embedding methods are highly efficient for real-time systems, while transformer-based embedding’s provide superior performance in complex semantic scenarios. These findings confirm the necessity of a flexible, multi-stage approach to the automated processing of public appeals.