Evolution of dimensions, metrics, and AI-enabled practices in a systematic mapping of data quality in the public sector

Data quality plays a critical role in ensuring the reliability and effectiveness of information systems in public sector organizations. However, fragmented data ecosystems, legacy systems, and the lack of standardized assessment practices hinder the development of consistent strategies. This study presents a systematic mapping of data quality dimensions, metrics, practices, and challenges across public institutions, following the guidelines proposed by Petersen et al. for systematic mapping studies. The work builds upon a previous mapping by the authors, expanding the study corpus from 53 to 63 peer-reviewed papers sourced from the IEEE, ACM, and Scopus databases. Completeness and Consistency remain the most frequently reported dimensions, cited in 44 and 42 of the 63 studies, respectively. Using a three-class coding scheme to distinguish actual from prospective Artificial Intelligence (AI) adoption, only 11 studies (17.46%) were found to explicitly implement AI or Machine Learning (ML) techniques for tasks such as anomaly detection and data imputation, whereas effective AI/ML adoption remained limited across the reviewed literature. The extended analysis further incorporates ontology-based data quality, AI governance, and an interpretive discussion of how the emergence of Large Language Models is reshaping data quality priorities. Persistent barriers — including automation gaps, governance deficits, and resource scarcity — are identified, informing a practical roadmap for data quality management to be applied at the Ceará State Treasury (Sefaz-CE).