Abstract
Software development has been plagued with the lack of documentation. We focus on ways to augment existing code comments with various external data sources such as a developer’s monologue or facial expression, allowing its user to track the accompanying non-textual information from a final source code. We propose CodeCAM, a framework to capture a developer’s reaction as well as its complete screenshot during a coding session as a video stream. A novel method is introduced to associate these streams with the corresponding portion of a source code. Our method does not require modifying existing tools or IDEs. We then applied facial expression analysis in attempt to capture a developer’s sentiment toward the source code during its development. Our preliminary experiments revealed that a developer tends to make a certain type of face (e.g. puzzled) when dealing with a difficult part of a program.