Recognizing Relevant Code Elements During Change Task Navigation

Developers spend a significant amount of their time exploring source code. Yet, little is known about the way developers break down their code exploration or the fine-grained navigation for change tasks within methods. The objective of our research is to address this gap and learn more about developers' code navigation for change tasks to devise better tool support. For our research, we perform exploratory studies also taking advantage of eye-tracking technology and interaction monitoring to gather detailed data on developers' code navigation down to the line-level. Based on the findings, we devise a model and approach that capture developers' code navigation and allow to automatically determine code elements that are relevant in the near future as well as to determine and summarize the elements that are currently relevant and might, for instance, be helpful for task resumption after interruptions. We plan to evaluate our model and approach in a multiple cases study on navigation recommendation and work resumption.