REST web service description for graph-based service discovery

Abstract

REST Web services have achieved momentum since they allow massive scalability, short response time, high availability and low entry barrier, among others. Unlike WSDL/SOAP based services, REST services lack a widely accepted and used service description since it increases the coupling between clients and servers, hampering service evolution. In practice, REST services are described through informal, ad-hoc and semi-structured descriptions, often written in natural language, which worsens the level of coupling. Most of the few REST service descriptions currently proposed follow an operation-centric approach with unclear additional benefits for developers and consumers. We propose a service description model focused on hypermedia that allows the generation of a graph that captures state transitions (i.e. HTTP operations) in an activity layer; we also capture resource, transition, and response semantics in a semantic layer. Using graph queries we traverse the graph and facilitate service discovery and composition. The service model was implemented as Microdata-based annotations, and a JSON description. A prototype was developed using Neo4J, and a set of real Web APIs was chosen to illustrate our approach.