[Paper Review] Feature Model-to-Ontology for SPL Application Realisation

Iis Afriyanti, Faiq Miftakhul Falakh+2 more|arXiv (Cornell University)|Jul 9, 2017

Semantic Web and Ontologies3 references3 citations

TL;DR

This paper proposes a method to transform a Software Product Line (SPL) feature model into an OWL ontology and use it to automatically generate a web application via the Zotonic framework. By mapping features, constraints, and business logic from the ontology to Zotonic’s category and predicate system, the approach enables end-to-end SPL application realization, demonstrating that ontologies can bridge problem-space modeling and solution-space implementation with linked data potential.

ABSTRACT

Feature model are widely used to capture commonalities and variabilities of artefacts in Software Product Line (SPL). Several studies have discussed the formal representation of feature diagram using ontologies with different styles of mapping. However, they still focused on the ontology approach for problem space and keep the solution space aside. In this paper, we present the modelling of feature model using OWL ontology and produce an application based on the ontology. Firstly, we map the features in a running example feature diagram to OWL classes and properties. Secondly, we verify the consistency of the OWL ontology by using reasoning engines. Finally, we use the ontology as an input of Zotonic framework for application realisation.

Motivation & Objective

• To close the gap between feature model modeling (problem space) and application realization (solution space) in SPL engineering.
• To enable machine-processable, semantically rich representation of SPL features using OWL ontologies.
• To demonstrate end-to-end application generation from feature models via ontology-driven framework integration.
• To support reuse, consistency checking, and linked data exposure through standardized ontology and web service integration.

Proposed method

• Mapping feature diagram elements (features, constraints, properties) to OWL classes, object properties, and datatype properties using a one-to-one transformation.
• Using reasoning engines to verify consistency of the generated OWL ontology.
• Integrating the OWL ontology as input into the Zotonic web framework for application generation.
• Mapping OWL classes to Zotonic categories and object properties to Zotonic predicates for structural alignment.
• Implementing business logic by detecting datatype property names matching predefined triggers, enabling automatic value computation.
• Executing a two-stage script process: first building the site, then running a mapper to inject ontology-defined structure into Zotonic’s admin and user interface templates.

Experimental results

Research questions

• RQ1 Can a feature model be formally represented as an OWL ontology to support both consistency checking and application realization?
• RQ2 How can constraints and relationships in a feature diagram be semantically mapped to an ontology and then to a web application framework?
• RQ3 To what extent can an ontology-driven approach enable automated, reusable, and semantically interoperable SPL application generation?
• RQ4 What is the feasibility of using the same ontology for both problem-space modeling and solution-space implementation in SPL?

Key findings

• The proposed method successfully mapped an AISCO feature model into an OWL ontology and generated a fully functional web application using the Zotonic framework.
• The ontology-enabled application supports dynamic form generation for datatype properties and integrates business logic based on property name patterns.
• Object properties in OWL were successfully translated into Zotonic predicates, enabling semantic relationships between application entities.
• The system demonstrated that OWL ontologies can serve dual roles: formal verification of feature models and direct input for application generation.
• The generated application exposes data via web services, and each feature is assigned a URI, enabling alignment with external ontologies like ORG and supporting linked data publishing.