Social Community Evolution Analysis and Visualization in Open Source Software Projects.

The importance of social communities around open-source software projects has been recognized. Despite that a lot of relevant research focusing on this topic, understanding the structures and dynamics of communities around open-source software projects remains a tedious and challenging task. As a result, an easily accessible and useful application that enables project developers to gain awareness of the status and development of the project communities is desirable. In this paper, we present MyCommunity, a web-based online application system to automatically extract communication-based community structures from social coding platforms such as GitHub. Based on the detected community structures, the system analyzes and visualizes the community evolution history of a project with a set of semantic-rich events, and quantify the strength of community evolution with respect to different events with a series of indexes. Built-in support to quantitative analysis and machine learning tasks based on the quantitative evolutionary events are provided. We demonstrate the usefulness of the system by presenting its ability in predicting project success or failure with the community evolution features. The results suggest the system achieves a prediction accuracy of 88.5% with commonly available machine learning models.