SHACL-DS: A SHACL extension to validate RDF dataset

Abstract: The Shapes Constraint Language (SHACL) provides a powerful mechanism for validating RDF data against shape constraints, but is inherently designed for single-graph validation. This limitation makes SHACL unsuitable for natively validating RDF datasets comprising multiple named graphs. To address this gap, developers must build solutions on top of SHACL, applying a shapes graph to each RDF dataset or combinations thereof using bespoke code. However, these approaches may lead to information loss, such as the named graph from which the data originates. This paper introduces SHACL-DS, an extension to SHACL that enables validation of RDF datasets. The extension adds a layer on top of SHACL, and the only disruptive change is the execution of SPARQL queries in, e.g., SPARQL-based constraints. The contributions are a SHACL-DS specification, a prototype implementation, and a set of test cases illustrating its use and providing future developers guidance in building SHACL-DS engines. This work lays the foundation for integrating dataset-level features into SHACL and encourages further exploration of advanced RDF dataset validation techniques.