The Violation Situation Pattern: Persistent Representation of Compliance Violations in Knowledge Graphs

Abstract:Existing compliance pipelines identify violations as transient query results, leaving no persistent representation of the violation itself or its lifecycle, evidence, and audit history. We address this limitation with the Violation Situation Pattern (VSP), a reusable ontology pattern that extends the Situation pattern of Gangemi and Mika by modeling each detected violation as a persistent first-class graph entity. Each violation is associated with a rule identifier, temporal validity interval, lifecycle state, and evidence links to the affected entities, while immutable lifecycle events provide a complete and queryable audit trail. We instantiate VSP on the schema of a deployed legal knowledge graph integrating corporate governance and contract data, populated with public contract corpora containing violations with established ground truth, and operationalize four deontic compliance rules. Rather than evaluating detection algorithms, we evaluate the pattern's robustness under rule evolution. Results show that repeated rule execution creates no duplicate violations, modifications to rule definitions preserve complete audit histories, and expanding rule scope maintains violation identity and lifecycle continuity. A case study using 73 regulatory enforcement decisions further demonstrates that compliance rules can evolve to capture additional confirmed violations without compromising the integrity of stored identities, evidence, or audit histories. These findings show that VSP enables compliance knowledge graphs to evolve without sacrificing traceability, explainability, or historical consistency. The complete implementation, queries, SHACL shapes, and evaluation dataset are publicly available to support the reproducibility of the reported results.

Artificial Intelligence (cs.AI)

arXiv:2606.03326 [cs.AI]

(or arXiv:2606.03326v2 [cs.AI] for this version)

https://doi.org/10.48550/arXiv.2606.03326 Focus to learn more arXiv-issued DOI via DataCite

arXivLabs is a framework that allows collaborators to develop and share new arXiv features directly on our website.

Both individuals and organizations that work with arXivLabs have embraced and accepted our values of openness, community, excellence, and user data privacy. arXiv is committed to these values and only works with partners that adhere to them.

Have an idea for a project that will add value for arXiv's community? Learn more about arXivLabs.