Multimodal artificial intelligence priorities and analysis portal in citizen services

Transparent governance means a lot to public grievance management, yet currently much of the existing digital portals in the world remain doing manual classification, a process that slows down the operations and directs a complaint to the inappropriate locations. This project is an introduction to Civic Connect, an AI-based multimodal grievance analysis and prioritization system based on BERT- (Bidirectional Encoder Representations from Transformers) and NLP with the goal to automate the handling of the complaints. The system allows citizens to report their complaints using a free text interface and voice notes or a custom built geo-tagged camera used to add the location and time data to the photos. On the back-end, the system intelligently classifies the complaints to either the proper departments and priorities based on an Explainable Prioritization Engine (X-PE) to make it obvious why an instance is urgent. To maintain fairness, it also has a Grievance Fairness Audit System (GFAS) to check for bias in the region as well as a Proactive Grievance Hotspot Prediction (PGHP) module that predicts spikes in complaints using historic data. It can be used as a web app with a Node.js/express backend and a Next.js dashboard to provide admins with real-time insights and visualizations, and this has made public administration more answerable and responsive than never before.