AI-Driven Smart Citizen Complaint Platform using Arabic NLP and Clustering

The Smart Citizen Complaint Platform is an AI-driven complaint management system designed to enhance the efficiency of collecting, analyzing, prioritizing, and routing citizen complaints. The platform integrates Arabic Natural Language Processing (NLP), complaint clustering, and multi-factor reliability scoring to reduce manual workload and support faster and more accurate decision-making in public service environments. The system accepts complaint text along with metadata such as category, area, date, and location. After secure OTP-based user verification, the platform applies transformer-based Arabic NLP models to automatically classify complaints and estimate their urgency The system architecture is role-based, involving citizens, employees, and administrators. Citizens can submit and track complaints, employees handle and update complaint status, and administrators monitor reports, system statistics, and operational performance through dynamic dashboards and interactive heat maps. An experimental evaluation was conducted using Arabic complaint datasets to assess the performance of transformer-based models. The results showed that AraBERT achieved the best performance with an accuracy of 0.9167 and an F1-score of 0.9167, outperforming MARBERT (accuracy 0.910, F1-score 0.909). The experiments also revealed that minimal preprocessing combined with AraBERTPreprocessor provides the most effective setup, while aggressive preprocessing techniques negatively impact performance due to loss of contextual information. This work was conducted at Arab International University (AIU), Syria. The official website of the university is: https://www.aiu.edu.sy