System and Preparation
We implemented the methodology described in Section III as a prototype.
The system consumes a set of URLs passed as an input file and fully automatically produces a Swagger file.
The input and output processing components are written in JavaScript, and the feature extraction and classification code is written in Python, relying on the scikit-learn library [9].
Communication between the components is achieved through the ZeroMQ messaging library.