Automating techniques for inspecting high assurance systems

Software inspections are an example of a validation technique for improving software product quality and reducing development time and cost. They are a peer-review process that have shown to decrease costs, decrease development time, and increase quality through a detailed examination of work-in-progress with the objective of identifying defects. In this paper, we report a methodology used for modeling the defects found during a code inspection. The Orthogonal Defect Classification (ODC) was used to categorize and make inferences that resulted in synthesized checklists that reflect the latest project experience and the rules most frequently broken during software development. We were then able to develop techniques that automatically used the checklists to search for defects in the source code. This automated defect detection technique will free up resources that can be used to look for more project specific issues.