An Indicator-Centric Knowledge Graph Model for Semantic Integration and Traceability of Official Statistical Indicators

Official statistical indicators are essential for monitoring progress toward the Sustainable Development Goals (SDGs). However, indicator data, metadata, source datasets, producing agencies, and global SDG references are often managed and disseminated in separate structures, limiting semantic integration, traceability, and machine-readability. This study proposes an indicator-centric knowledge graph model for integrating and tracing SDG indicators in Indonesia. The model positions StatisticalIndicator as the central entity that connects SDGs, SDG targets, global indicator references, national and proxy indicator metadata, source datasets, publishing agencies, controlled codelists, and statistical observations. Following a reuse-first design principle, the model was implemented using established semantic web vocabularies, extended with only seven custom properties. The model is instantiated and evaluated through a purposively selected case of 16 child-related SDG indicators in Indonesia, chosen to maximize structural heterogeneity across producing agencies, data sources, and indicator–framework relationships, including nationally adapted and proxy indicators. The knowledge graph was evaluated using SPARQL-based competency questions with predefined ground truth and query execution time measurement, SHACL structural validation complemented by mutation-based negative testing, traceability coverage analysis with source verification, and supplementary expert judgement. The findings demonstrate that the proposed indicator-centric knowledge graph model can support semantic integration, metadata access, and end-to-end traceability of official statistical indicators with a minimal, interoperable vocabulary extension. While the empirical evidence is limited to a single-country, single-domain instantiation, the model itself is domain-agnostic; its layers and relations are defined independently of the thematic content of the indicators.