TLINGA: A Local-First Urban Intelligence Framework for Social Media Signal-to-Action Recommendation

Abstract

Citizen aspirations are important inputs for responsive local governance, yet formal complaint channels require structured reporting, while conventional social-media monitoring often focuses only on sentiment and trends. This paper proposes TLINGA, a Local-First Urban Intelligence Framework that transforms naturally expressed comments in locality-bound government social-media spaces into structured decision-support inputs. The framework is demonstrated through a multi-jurisdictional case study involving one provincial and five city/regency governments. TLINGA consists of six stages: Track identifies relevant local sources; Listen collects public comments; Integrate consolidates records into a unified database; Navigate extracts locality and contextual information; Guide categorizes issues, assigns responsible agencies, and prioritize cases; and Act generates intervention and follow-up recommendations, including potential planning or budget pathways. The implementation combines automated data acquisition, standardized storage, language-model-based context extraction, spatial mapping, routing rules, and human validation. Rather than automatically executing interventions or allocating funds, TLINGA produces an auditable recommendation package containing the citizen signal, inferred location, issue category, responsible agency, priority level, and suggested action. The framework offers a practical approach for converting locality-bound citizen discourse into actionable urban intelligence.