Monitoring model analytics over large repositories with Hawk and MEASURE

Abstract Software analytics play an important role in developing systems today. They can provide continuous and valuable insights into areas of interest throughout the life cycle of a system, guiding the engineers to the areas that need to be worked on to create timely and robust software. An important requirement is to have a low-effort way of producing them, without largely affecting current work patterns. This work presents the integration of the Hawk model indexing framework into the MEASURE metrics platform, allowing for the efficient production of model metrics over large collections of models managed by teams of developers. We evaluate this integration by monitoring a real-world model with hundreds of changes over its lifetime and producing relevant metrics over this period of time. This evaluation has demonstrated the ability to produce complex metrics and evaluate them over the entire history of a model.