A research-stage framework uses symbolic AI, formalized regulations, SPARQL and an ontology to generate justified compliance suggestions for a human user. When reasoning fails because rules are inconsistent or the situation is insufficiently described, an explainability module identifies the problem and either requests missing information or reports the inconsistency.
The system distinguishes decided from undecided outcomes. Decided results include permission, facultative, obligation and prohibition. Undecided results are either indecisive, when no formal rule is triggered, or contradictory, when applicable rules produce conflicting deontic answers.
For a decided case, the framework presents the conclusion with the legal sources of the rules that produced it. For a contradictory case, it identifies the conflicting rules, asks which should prevail, records a temporary conflict-solving rule and forwards it for expert validation. For an indecisive case, it asks the user for information that could support a conclusion.
The approach has material limits. Rule extraction is currently manual, and the regulations used must be adapted to the scenario. The demonstrated use case concerns data sharing and processing by European Law Enforcement Agencies, while the authors describe the tests as proof-of-concepts and identify the need for a broader rule set and a full benchmark using real-world data.
For enterprise governance teams, the supported pattern is human-reviewed decision support; formalize the applicable rules, show the sources behind decided outcomes, and route incomplete or conflicting cases for additional information or expert review. The research does not establish that this framework is ready to determine compliance across broader enterprise contexts.