Generation of an architecture view for web applications using a bayesian network classifier

A recurring problem in software engineering is the correct definition and enforcement of an architecture that can guide the development and maintenance processes of software systems. This is due in part to a lack of correct definition and maintenance of architectural documentation. In this paper, an approach based on a bayesian network classifier is proposed to aid in the generation of an architecture view for web applications developed according to the Model View Controller (MVC) architectural pattern. This view is comprised of the system components, their inter-project relations and their classification according to the MVC pattern. The generated view can then be used as part of the system documentation to help enforce the original architectural intent when changes are applied to the system. Finally, an implementation of this approach is presented for Java based-systems, using training data from popular web development frameworks.