Using Knowledge Graph Technologies to Contextualize and Validate Declarations in the Social Security Domain

Abstract

This master’s thesis explores the feasibility and potential of knowledge graph technologies, with a specific emphasis on the Shapes Constraint Language (SHACL), for validating declaration forms in the Belgian Social Security domain. The study focused on validating DmfA declarations which are reports describing the work of employees done during a quarter sent by an employer to the government. The study first created a vocabulary for a knowledge graph based on DmfA declarations and the mapping of diverse data sources into Resource Description Framework (RDF) format. The heart of the work lies in the declaration of SHACL constraints for complex business rules, many of which were implemented via SPARQL, highlighting SHACL’s scalability and versatility in addressing challenging validation tasks. Despite obstacles concerning data transformation while generating RDF representations of the declarations, the research demonstrates that SHACL provides a richer application profile and exceeds the expressiveness of XML Schema Definition (XSD) constraints in implementing complex validation rules. These findings illustrate the potential of knowledge graph technologies and SHACL for managing intricate validation tasks in social security systems and hint at the prospect of their application in future work, such as evolving document integration and validation report integration using named graphs and provenance information.