UpSHACL: Targeted Constraint Validation for Updates over Knowledge Graphs

Zenon G. Zacouris, Jin Ke, Maribel Acosta

Abstract: Knowledge Graphs (KGs) evolve frequently through updates that reflect new information or corrections, but such changes can compromise data quality constraints. For KGs modelled in RDF, such constraints can be expressed with the Shapes Constraint Language (SHACL). Current SHACL engines are tailored to validating static graphs but are inefficient for KGs under updates, as they require re-validating the full graph after each update. In this paper, we present UpSHACL, an approach for efficient SHACL validation after updates. UpSHACL introduces a formal model to identify the subgraph affected by an update and constructs a reduced subgraph that can be validated using existing SHACL engines. Our algorithm, implemented with SPARQL over RDF triple stores, integrates seamlessly with Semantic Web technologies. Experimental results show that UpSHACL achieves up to 10\(\times \) speedup over static full validation, with performance gains increasing on larger KGs.

External IDs:dblp:conf/semweb/ZacourisKA25