Increasing the Responsiveness of Recommended Expert Collaborators for Online Open Projects
Mohammad Y. Allaho, Wang-Chien Lee · 2014
We consider the experts recommendation problem for open collaborative projects in large-scale Open Source Software (OSS) communities. In large-scale online community, recommending expert collaborators to a project coordinator or lead developer has two prominent challenges: (i) the "cold shoulder"' problem, which is the lack of interest from the experts to collaborate and share their skills, and (ii) the "cold start" problem, which is an issue with community members who has scarce data history. In this paper, we consider the Degree of Knowledge (DoK) which imposes the knowledge of the skills factor, and the Social Relative Importance (SRI) which imposes the social distance factor to tackle the aforementioned challenges. We propose four DoK models and integrate them with three SRI methods under our proposed Expert Ranking (ER) framework to rank the candidate expert collaborators based on their likelihood of collaborating in response to a query formulated by the social network of a query initiator and certain required skills to a project/task. We evaluate our proposal using a dataset collected from Github.com, which is one of the most fast-growing, large-scale online OSS community. In addition, we test the models under different data scarcity levels. The experiment shows promising results of recommending expert collaborators who tend to make real collaborations to projects.