Representing Normative Regulations in OWL DL for Automated Compliance Checking Supported by Text Annotation
Ildar Baimuratov, Denis Turygin
TL;DR
The work tackles the challenge of automating compliance checking for normative regulations by proposing a text annotation schema tightly aligned with OWL DL and a recursive algorithm that transforms annotated text into machine-interpretable OWL DL code. This enables reasoning-based compliance verification in domains like building construction, addressing limitations of SHACL and ML-only approaches by providing explainable, traceable, and auditable results. A proof-of-concept demonstrates end-to-end conversion and reasoning on annotated regulations (including qualitative and quantitative examples), validated with a Pellet reasoner and a Python-based prototype. The approach yields a transparent workflow where domain experts control semantic modeling while machine assistance supports annotation, aiming for trustworthy automated regulation checking with practical impact for complex regulatory codes. The work highlights both the potential and current limitations (e.g., vocabulary coverage, domain gaps) and outlines future directions toward ML-assisted annotation without relinquishing human oversight.
Abstract
Compliance checking is the process of determining whether a regulated entity adheres to these regulations. Currently, compliance checking is predominantly manual, requiring significant time and highly skilled experts, while still being prone to errors caused by the human factor. Various approaches have been explored to automate compliance checking, however, representing regulations in OWL DL language which enables compliance checking through OWL reasoning has not been adopted. In this work, we propose an annotation schema and an algorithm that transforms text annotations into machine-interpretable OWL DL code. The proposed approach is validated through a proof-of-concept implementation applied to examples from the building construction domain.