The Design And Implementation Of An Expert Recommendation System For Source Code

When performing tasks in software maintenance phase, software engineers often encounter unfamiliar source code. When they need to understand the design idea and related information of this kind of source code, they will actively seek cooperation from experienced colleagues. The colleagues are regarded as source code experts.During the process of seeking collaboration, the engineers often can not get enough information to judge whether a colleague is a source code expert. This problem causes the action they take afterward less reasonable. And these actions often cause negative effect on them and their colleagues.The final result is the low efficiency of the product maintenance tasks. So, it is a valuable research topic that how to provide enough valuable information to the engineers to support them to make reasonable judgment.For a certain source code fragment, to the engineer who is not the author, its visible time information generated when the code is visible to him is valuable for the judgment of the possibility that he has the knowledge related to its design idea. Based on the idea, this thesis proposed a general design idea of a kind of source code expert recommendation system, by utilizing the personal source code comprehension time information, the author information provided by the source code version control system and the corresponding feedbacks provided by previous cooperation seekers.Based on the design idea, for the java software engineers using Eclipse IDE, the thesis designed and implemented a corresponding source code expert recommendation system. The main functions of the system are recording personal source code comprehension time information, collecting basic potential expert information, providing potential expert information. It can effectively help the engineers to seek source code experts by generating a form of potential expert information according to the source code provided by them.