The market of data-intensive Application Programming Interfaces (APIs) has recently experienced an exponential growth, but the creation and evolution of such APIs is still done ad-hoc, with little automated support and reported deficiencies. These drawbacks hinder the productivity of developers of those APIs and the services built on top of them. In this exploratory paper, we promote a data-driven approach to improve the automatization of data-intensive API creation and evolution. In a release cycle, data coming from API usage and developers will be gathered to compute several indicators whose analysis will guide the planning of the next release. This data will also help to generate complete documentation facilitating APIs adoption by third parties.

Peer Reviewed