AI-Assisted Civic Feedback Engine

The AI-Assisted Civic Feedback Engine is a research-driven pilot that applies modern natural-language processing to thousands of citizen comments, complaints, and suggestions collected through municipal channels.

Objectives

• Train a multilingual (EL/EN) sentiment & topic-classification model on real municipal data
• Build a dashboard that clusters feedback by neighbourhood, urgency, and department
• Produce a policy brief with data-backed recommendations for service improvements

Methodology

The project follows a human-in-the-loop approach: automated classification is reviewed by civil-servant annotators, ensuring accuracy while continuously improving the model. An ethics review board oversees bias auditing at every training cycle.

Expected Outcomes

Reduction of median response time from 12 days to under 4 days for priority-flagged issues, and a 55 % decrease in mis-routed citizen tickets.