Ontology-Driven Model-to-Model Transformation of Workflow Specifications

TL;DR

This paper presents an ontology-driven model-to-model pipeline that translates proprietary JSON-based workflow definitions into BPMN 2.0 diagrams. The approach separates mapping knowledge into ontologies and declarative rules, then lifts data to RDF/OWL, aligns it with a BPMN ontology through reasoning, and generates standard BPMN diagrams using the Camunda API, with deterministic layout and batch processing suitable for CI. In a IST case study with 69 workflows, the pipeline produced 92 BPMN diagrams with a 94.2% success rate and an average translation time of about 404 ms per file, while qualitative interviews reported improved comprehension, onboarding, and traceability. The work demonstrates that ontology-driven M2M transformations can bridge proprietary workflow formats and open standards, offering interoperability, maintainability, and a path toward adapter-based generalization across vendors and languages. The contribution advances semantic workflow engineering by externalizing mappings, preserving traceability from source definitions to BPMN outputs, and enabling repeatable, scalable translations in enterprise settings.

Abstract

Proprietary workflow modeling languages such as Smart Forms & Smart Flow hamper interoperability and reuse because they lock process knowledge into closed formats. To address this vendor lock-in and ease migration to open standards, we introduce an ontology-driven model-to-model pipeline that systematically translates domain-specific workflow definitions to Business Process Model and Notation (BPMN) 2.0. The pipeline comprises three phases: RML-based semantic lifting of JSON to RDF/OWL, ontology alignment and reasoning, and BPMN generation via the Camunda Model API. By externalizing mapping knowledge into ontologies and declarative rules rather than code, the approach supports reusability across vendor-specific formats and preserves semantic traceability between source definitions and target BPMN models. We instantiated the pipeline for Instituto Superior Técnico (IST)'s Smart Forms & Smart Flow and implemented a converter that produces standard-compliant BPMN diagrams. Evaluation on a corpus of 69 real-world workflows produced 92 BPMN diagrams with a 94.2% success rate. Failures (5.81%) stemmed from dynamic behaviors and time-based transitions not explicit in the static JSON. Interviews with support and development teams indicated that the resulting diagrams provide a top-down view that improves comprehension, diagnosis and onboarding by exposing implicit control flow and linking tasks and forms back to their sources. The pipeline is generalizable to other proprietary workflow languages by adapting the ontology and mappings, enabling interoperability and reducing vendor dependency while supporting continuous integration and long-term maintainability. The presented case study demonstrates that ontology-driven M2M transformation can systematically bridge domain-specific workflows and standard notations, offering quantifiable performance and qualitative benefits for stakeholders.

Figures (2)

• Figure 1: Component structure and artifact relationships in the ontology-driven translation pipeline.
• Figure 2: Runtime data flow from input selection through lifting, reasoning, generation, layout and persistence.