A practical approach to services composition through light semantic descriptions

Services composition has been much investigated over the last decade without reaching shared and consolidated results mainly for the lack of interoperable descriptions of services and the consequent need of extensive user intervention. In this paper, we propose a light and practical approach to create machine-readable descriptions of output data that can be merged or used (as-is or adapted) as input data to other services. The solution relies on the popular and standard OpenAPI descriptions augmented with annotations based on JSON-LD format. Services descriptions are created by table annotations techniques applied on sets of given or retrieved output values. The approach has been implemented in a tool and validated with a set of real services.