ABSTRACT
CSIRO Computational Informatics, Australian National University, Australia
Information services are commonly provided via Web APIs based on Representational State Transfer (REST) principles [196,452] or via Web Services based on the WS-* technology stack [182,429]. Currently deployed information services use HTTP as transport protocol, but return data as JSON or XML which requires glue code to combine data from different APIs with information provided as Linked Data. Linked Data interfaces for services have been created, e.g., in form of the book mashup [97] which returns RDF data about books based on Amazon’s API, or twitter2foaf which encodes the Twitter follower network of a given user based on the API provided by Twitter. However, the interfaces are not formally described and thus the link between services and data has to be established manually or by service-specific algorithms. For example, to establish a link between person instances (e.g., described using the FOAF vocabulary1) and their Twitter account, one has to hard-code which property relates people to their Twitter username and the fact that the URI of the person’s Twitter representation is created by appending the username to https://twitter2foaf.appspot.com/id/.
