AI Use Case: Measuring Representativeness of Civic Datasets Against a Declared Reference Population

Abstract

Open contribution to ISO/IEC JTC 1/SC 42 (candidate input to ISO/IEC TR 24030) and the ISO/IEC 5259 series. Demonstrates on real public-sector data from Berlin and Bonn that civic datasets can score 100% on conventional data-quality dimensions while under-recording whole districts, and proposes five requirements for a normative representativeness procedure. Prepared under the StandICT.eu 2029 Fellowship, contract 02/71.