From RDF Graph Validation to RDF Dataset Validation with SHACL-DS

Abstract

The Shapes Constraint Language (SHACL) is the W3C Recommendation for validating a single RDF graph. This makes SHACL inadequate for validating data across (named) graphs in an RDF dataset. Existing workarounds, such as graph unions or bespoke preprocessing, either collapse the RDF dataset structure or compromise the declarative nature of SHACL validation. In the former, we lose track of where triples come from; in the latter, knowledge is hidden in the code, and the constraints are not self-contained nor fully declarative. We present SHACL-DS to address this problem. SHACL-DS proposes a vocabulary and an algorithm on top of SHACL for RDF dataset validation. SHACL-DS introduces the concepts of Shapes Datasets, Target Graph Declarations, and Target Graph Combinations, enabling declarative constraints to operate across multiple graphs in an RDF dataset. SHACL-DS also defines the behaviour of SPARQL-based constraints for validating RDF datasets. In this paper, we formalize SHACL-DS and provide a prototype implementation.