Assessing quality attributes of open source software by mining low ceremony evidence

Graduation date: 2013Programmers often have to choose components online for reuse based on software quality. To help with this choice, most component repositories (SourceForge, CodeProject, etc.) provide information such as user ratings and reviews of components. However, the reusability of components is not immediately obvious from\ud this material. To make things worse, reviews and other material could potentially contradict one another about reusability or any other issue. Moreover, there could be multiple components that claim to perform identical tasks in which case it becomes difficult to identify the most reusable component. This thesis presents the result of two studies aimed at discovering how to use material from a component repository to automatically characterize the reusability of components. It was found that two factors (out of three) were significant predictors of reusability. Additionally, a review summarizer was developed to summarize user reviews and return the overall opinion reflected by the comments