Software Metrics in Boa Large-Scale Software Mining Infrastructure: Challenges and Solutions

In this paper, we describe our experience implementing some of classic software engineering metrics using Boa—a large-scale software repository mining platform—and its dedicated language. We also aim to take an advantage of the Boa infrastructure to propose new software metrics and to characterize open source projects by software metrics to provide reference values of software metrics based on large number of open source projects. Presented software metrics, well known and proposed in this paper, can be used to build large-scale software defect prediction models. Additionally, we present the obstacles we met while developing metrics, and our analysis can be used to improve Boa in its future releases. The implemented metrics can also be used as a foundation for more complex explorations of open source projects and serve as a guide how to implement software metrics using Boa as the source code of the metrics is freely available to support reproducible research.