Data Mining for Software Engineering

Reviewer: Alexis Leon

Nowadays, almost every aspect of life is touched and controlled by software. As software becomes a more prominent presence, the task of developing it becomes more difficult. Because software is used for critical applications and for controlling sophisticated equipment and systems, even a small bug can potentially have catastrophic consequences. Today's software projects are becoming more complex in size, sophistication, and the technologies used. As demand for software increases, software developers must find ways to improve their productivity, efficiency, and the quality of their product. This paper explores the use of data mining algorithms to perform software engineering (SE) tasks, and improve the quality and productivity of software developers. Organizations have large amounts of SE data, such as documents, reports, source code, and bug reports; mining this data can help organizations solve many SE problems. Data mining algorithms can help software engineers find the correct usage of an application programming interface (API), the impact of a change in source code, and potential bugs in the software. This paper explains the mining technology, the various challenges of mining SE data, and common data mining algorithms. Using examples, Xie et al. explain the various data, graph, and text mining algorithms for mining SE data: iterative patterns, temporal rules, sequence diagrams, finite state machines, sequence association rules, discriminative graph mining, graph classification to assist in debugging, and an algorithm to detect duplicate bug reporting. The results from their experiments illustrate the superiority of these techniques. This excellent paper will be useful to software engineers, productivity improvement professionals, and people and organizations that develop tools for improving the quality and efficiency of software development. Online Computing Reviews Service