School of Engineering, University of Electronic Science and Technology of China, Chengdu 611731, China;Technology Department, Hunan Broadcasting System, Changsha 410007, China 在期刊界中查找 在百度中查找 在本站中查找
Information retrieval (IR) is widely used in automatically discovering requirement traceability. However IR will miss some correct artifacts which have low text similarity with the requirement. There are accuracy issues in requirement traceability based on IR. To solve the problem, we propose an approach of using the developer collaborative relationship to improve the accuracy of the traceability links recovery between requirement and source code. Meanwhile, we develop a requirement-to-code traceability system. When the system is tracing, it retrieves the source code artifacts of the highest text similarity with the requirement and extracts the developer collaborative relationship from code commit logs. Then the system recommends some relevant code artifacts by developer collaboration relationship. Users can choose the correct code artifacts from the recommend result. The experiment shows that the requirement traceability system could improve the accuracy and the efficiency and reduce errors.