ITMViz: Interactive Topic Modeling for Source Code Analysis

Topic modeling has seen a surge in use for softwarecomprehension. Although the models inferred from the sourcecode are a great source of knowledge, they fail to fully capture theconceptual relationships between the topics. Here we investigatethe use of interactive topic modeling for source code analysis byfeeding-in information from the end-users, including developersand architects, to refine the inferred topic models. We haveimplemented a web-based toolkit called ITMViz to providesupport to interpret the topic models, and use the results tocluster modules together. A medium-sized Java project is usedto evaluate our approach in understanding the software system.