The LRA Workbench: An IDE for Efficient REST API Composition through Linked Metadata

Abstract

The number of Web APIs for accessing information and services is continuously increasing, and yet, no tools exist to automate the time-consuming and error-prone process of invoking these APIs and composing their responses. The recent emergence of widely-adopted, standardized, Web-API description formats and the development of Linked Data technologies for data integration have motivated our work on the LRA (Linked REST APIs) methodology. LRA relies on RDF service specifications to automate the development process around the usage of Web APIs. This automation represents a great opportunity to systematize and improve the quality of service-oriented application development. However, LRA's reliance on SPARQL as the user-interaction model may hinder its adoption, because it requires developers to learn the intricacies of the unconventional graph data model and its associated datasets. In this paper we have developed the LRA Workbench (LRAWbench), which takes advantage of the emergent schema of Web-API specifications, in order to simplify the formulation of LRA-compliant SPARQL queries. Our empirical evaluation of the LRAWbench usability demonstrates that our tool significantly improves the performance of developers formulating SPARQL queries for LRA. A subsequent study on the effectiveness of the LRAWbench demonstrated that developers using LRA tend to produce code with considerable better structural complexity, in less time, than developers manually composing APIs.