A Multiscale Geospatial Community Intelligence Architecture for AI-Assisted Urban Incident Management: The Alartia Platform

Abstract

Urban community safety increasingly depends on transforming citizen-generated information into structured, traceable, and territorially meaningful data, particularly where neighborhood coordination relies on informal communication channels with limited geolocation and institutional integration. This study designs and implements Alartia, a multiscale community intelligence architecture for AI-assisted urban incident management that connects resident reporting, community administration, geospatial processing, and municipal decision support. Using Design Science Research with iterative Scrum development, the architecture combines GeoJSON/PostGIS for exact territorial representation, H3 4.4.2 for multiresolution indexing and aggregation, an IRIS incident admission pipeline integrating deterministic preprocessing with DeepSeek-assisted moderation, cloud-native deployment through Google Cloud Platform and Terraform, and a prototype patrol prioritization mechanism based on severity-derived risk, temporal decay, and geodesic proximity. The resulting artifact integrates persistent incident traceability, multiscale territorial intelligence, AI-assisted moderation, and prototype municipal support. Pilot-informed refinement introduced low-friction SOS reporting, accessibility-oriented adjustments, and expanded municipal analytics. The results distinguish implemented artifact outcomes, analytical/model-based prototype behavior, and scenario-based quantitative evidence, thereby separating technological realization from claims of operational effectiveness in real settings. Techno-economic analysis characterizes infrastructure behavior under explicit workload, pricing, and adoption assumptions rather than observed production conditions. Overall, Alartia demonstrates architectural integration and implementation-level traceability, while predictive accuracy, production-scale performance, operational dispatch superiority, usability, adoption, and field-level public safety effectiveness remain subjects for subsequent empirical evaluation.