Constraints to Validate RDF Data Quality on Common Vocabularies in the Social, Behavioral, and Economic Sciences
To ensure high quality of and trust in both metadata and data, their representation in RDF must satisfy certain criteria - specified in terms of RDF constraints. From 2012 to 2015 together with other Linked Data community members and experts from the social, behavioral, and economic sciences (SBE), we developed diverse vocabularies to represent SBE metadata and rectangular data in RDF. The DDI-RDF Discovery Vocabulary (DDI-RDF) is designed to support the dissemination, management, and reuse of unit-record data, i.e., data about individuals, households, and businesses, collected in form of responses to studies and archived for research purposes. The RDF Data Cube Vocabulary (QB) is a W3C recommendation for expressing data cubes, i.e. multi-dimensional aggregate data and its metadata. Physical Data Description (PHDD) is a vocabulary to model data in rectangular format, i.e., tabular data. The data could either be represented in records with character-separated values (CSV) or fixed length. The Simple Knowledge Organization System (SKOS) is a vocabulary to build knowledge organization systems such as thesauri, classification schemes, and taxonomies. XKOS is a SKOS extension to describe formal statistical classifications. In this paper, we describe RDF constraints to validate metadata on unit-record data (DDI-RDF), aggregated data (QB), thesauri (SKOS), and statistical classifications (XKOS) and to validate tabular data (PHDD) - all of them represented in RDF. We classified these constraints according to the severity of occurring constraint violations. This technical report is updated continuously as modifying, adding, and deleting constraints remains ongoing work.