AUTHOR=Nicholson Nicholas , Štotl Iztok TITLE=A generic framework for the semantic contextualization of indicators JOURNAL=Frontiers in Computer Science VOLUME=6 YEAR=2024 URL=https://www.frontiersin.org/journals/computer-science/articles/10.3389/fcomp.2024.1463989 DOI=10.3389/fcomp.2024.1463989 ISSN=2624-9898 ABSTRACT=

Indicators are quantitative or qualitative measures used to gauge various aspects of society and assess change over time (such as monitoring the progress or effectiveness of a public policy). Ideally, indicators should be precisely defined and measured according to harmonized procedures that may not be feasible in practice, especially in domains such as health, where indicators are often derived from preexisting, heterogeneous datasets. Integrating such data has posed a persistent challenge, but semantic technologies offer advantages by enriching data in a relatively simple, linkable, and non-disruptive way. However, without harmonized frameworks, the difficulties associated with data integration are unlikely to be resolved. In this article, we propose a generic, domain-neutral indicator contextualization framework for structuring and linking distributed datasets with contextual metadata according to a standard model. The framework integrates the concepts of the International Organization for Standardization/International Electrotechnical Commission (ISO/IEC) 11179 metadata registry standard with the common core ontologies (CCO) mid-level ontology suite, and incorporates other semantic technologies to make it adaptable and interoperable within and across domains. Application of the framework to an example indicator illustrates the versatility and adaptability of the approach in a federated data architecture. The contextual information can be dereferenced using standard query tools to provide data users a comprehensive understanding and overview of the indicator. The framework is amenable to deep learning applications via the principles of semantic data models, linked open data, and knowledge organization systems. The ideas are presented to stimulate further reflection and consolidation of standard data contextualization frameworks.