Towards Generation of Software Development Tasks

Abstract

The presence of well defined fine-grained sub-tasks is important to the development process: having a fine-grained task context has been shown to allow developers to more efficiently resume work. However, determining how to break a high level task down into sub-tasks is not always straightforward. Sometimes developers lack experience, and at other times, the task definition is not clear enough to afford confident decomposition. In my research I intend to show that by using syntactic mining of past task descriptions and their decomposition, I can provide automatically derived sub-task suggestions to afford more confident task decomposition by developers.