Recommending software experts using code similarity and social heuristics
Ghadeer A. Kintab, Chanchal Kumar Roy, Gordon I. McCalla · 2014
Successful collaboration among developers is crucial to the completion of software projects in a Distributed Software System Development (DSSD) environment. We have developed an Ex-pert Recommender System Framework (ERSF) that assists a developer (called the “Active Devel- oper”) to find other developers who can help them to fix code with which they are having difficulty. The ERSF first looks for other developers with similar technical expertise, as measured by their prior work on code fragments that are similar to (clones of) the code that the Active Developer is working on (the “code at hand”). As well, it ana-lyzes the other developers ’ social relationships with the Active Developer (available from the DSSD environment) and their social activities within the ERSF (information which helps to maintain developer profiles used in this analysis). This information is then combined to provide a ranked list of potential helpers based on both technical and social measures. A proof of concept experiment shows that the ERSF can recommend experts with good to excellent accuracy, when compared with human rankings of appropriate experts in the same scenarios 1