A Linked Data Model for Web API-s

Abstract

Web APIs (Application Programming Interface) provide means for rapid enterprise integration. The progress in Web API-s has reached the level of maturity where the impact of new advances can and should be validated in practical setting. Usage of various datasets have been reported in the literature of Web API-s. These datasets could allow systematic validation of methods such as composition, selection and recommendation, just to mention a few. However, we have seen that there are obstacles in using these datasets. More specifically, sometimes the datasets are not publicly available and the authors cannot be easily contacted. Furthermore, although some data is available in unprocessed format, the data processing steps are often not completely revealed to facilitate replication of experimental results. Finally, although there are multiple datasets available, there is no common metamodel, which would allow building a unified view to the data and thereby allow efficient development of Web API evaluation suites. In this paper we extend Linked USDL model to embrace a variety of available Web API datasets using linked data principles. Applicability of the metamodel is validated via encoding two Web service’s datasets, which are made available for community use.