Grant: Engage Grants Program

Software has become increasingly essential for nearly all areas of business - from inventory and billing to human resource management. These systems are often very complex, and require integration of new components with legacy components. Moreover, these components are created by many people, often with differing coding competencies and styles. Interdependencies of components and dynamic data compound the complexity of the problem. Finding code segments that generates soft (non-fatal) errors or that is unnecessarily slow is thus typically beyond human capacity because of the complexity. Therefore, there is a high-value opportunity to develop algorithms that can detect problematic code segments so they can be fixed. This increases the speed and reliability of the code for users, which increases business efficiency. CodeExcellence is a leading Canadian provider of quality-control service for custom code written for SAP, the largest business information management software company. CodeExcellence has developed proprietary metrics for automated quality-assurance that helps their customers improve the quality of their custom SAP code. These metrics are based on general heuristics, and do not currently use information about performance in the native environment. This ENGAGE project will exploit current methods in machine learning in order to use runtime information (errors, speed, ..) to identify the most detrimental coding bugs. This will be an important advance that will let users prioritize debugging, and thereby maximize coding time to increase performance. Addition of this technology will allow CodeExcellence to boost efficiency and performance while reducing maintenance costs for their customers. This will add value to CodeExcellence's services and increase their profitability, market share, and employment