Civicflow: a retrieval-augmented multimodal civic issue triage system for urban governance
Abstract
Urban municipalities in rapidly expanding cities process in excess of 10,000 multimodal civic complaints monthly, encompassing road damage, water supply failures, electricity outages, sanitation deficiencies, and public safety hazards. Existing complaint management systems are constrained by manual classification pipelines that are unable to process heterogeneous input modalities, are susceptible to systematic misrouting, and provide no mechanisms for real-time tracking or data-driven governance analytics. This paper proposes CivicFlow, a novel retrieval-augmented multimodal triage architecture that automates probabilistic classification and geo-aware routing of civic complaints. CivicFlow integrates Automatic Speech Recognition (ASR) via OpenAI Whisper [4], vision-language captioning via CLIP ViT-B/32 [3], dense semantic embeddings via Sentence-BERT [5], and Llama-3 8B Instruct [14] augmented with Retrieval-Augmented Generation (RAG) [2] over a municipal policy vector store. Experimental evaluation on 1,247 labeled Bangalore civic complaints demonstrates a macro-averaged F1 score of 87.3%, end-to-end latency of 2.41 s (P95), and duplicate detection precision of 91.2%. To the best of our knowledge, this is the first paper to propose a fully automated, multimodal, RAG-grounded civic complaint triage pipeline benchmarked on real Bangalore municipal data.
Read the paper
Related studies
- Knowledge Management Perspective of Generative Artificial Intelligence
- Navigating Generative Artificial Intelligence Promises and Perils for Knowledge and Creative Work
- Characterizing the evolving research on enterprise content management
- Large language models and multimodal foundation models for precision oncology
- From Generative AI to Extended Reality: Multidisciplinary Perspectives on the Challenges, Opportunities, and Future of Educational Computing
- Generative Artificial Intelligence: A New Engine for Advancing Environmental Science and Engineering
Go deeper in Evidence Hub
Evidence Hub adds a study data sheet (design, sample size, evidence level, key numbers), an AI summary and a chat grounded in the paper's text.