Automated Ontology Instantiation of OpenAPI REST Service Descriptions

Abstract

Web services are published in service registries on the Web by various software vendors and cloud providers so that can be easily discovered and used by the users. The need for standardizing technologies for service publishing and discovery is of crucial importance for their adoption and market success. Web services need to be described in a way that eliminates ambiguities so that they can be uniquely identified by users or machines. OpenAPI Specification is a powerful framework for the description of REST APIs. OpenAPI standard aims to provide service descriptions which are understandable by both humans and machines. Despite its rigorous service language format, OpenAPI service descriptions can be vague. In our previous work we identified the causes of ambiguity in service descriptions and showed that ambiguities can be eliminated by associating ambiguous OpenAPI properties to concepts in a semantic model. Each service is then uniquely described by a JSON object representing its identity, properties, purpose and functionality. Further to the above objective, the Semantic Web vision defines the requirements for unifying the world of Web services and suggests representing the services as semantic objects accessible on the Web. All services should be published on the Web by means of Semantic Web tools (e.g. ontologies), become discovered by Web search engines (e.g. SPARQL) and be reused in applications. Leveraging latest results for hypermedia-based construction of Web APIs (i.e. Hydra) and the latest update of the OpenAPI specification (OpenAPI v3.0), the present work proposes a reference ontology for REST services along with a formal procedure for converting OpenAPI service descriptions to instances of this ontology.