Focusing Automatic Code Inspections

Automatic Code Inspection tools help developers in early detection of defects in software. A well-known drawback of many automatic inspection approaches is that they yield too many warnings and require a clearer focus. In this thesis, we provide such focus by proposing two methods to prioritize inspection results, based on the execution likelihood and the historical fault likelihood of warnings. We describe both methods in detail, and evaluate them on various open-source and industry projects.