Personalized teammate recommendation for crowdsourced software developers.

Luting Ye,Hailong Sun,Xu Wang,Jiaruijue Wang
DOI: https://doi.org/10.1145/3238147.3240472
2018-01-01
Abstract:Most crowdsourced software development platforms adopt contest paradigm to solicit contributions from the community. To attain competitiveness in complex tasks, crowdsourced software developers often choose to work with others collaboratively. However, existing crowdsourcing platforms generally assume independent contributions from developers and do not provide effective support for team formation. Prior studies on team recommendation aim at optimizing task outcomes by recommending the most suitable team for a task instead of finding appropriate collaborators for a specific person. In this work, we are concerned with teammate recommendation for crowdsourcing developers. First, we present the results of an empirical study of Kaggle, which shows that developers’personal teammate preferences are mainly affected by three factors. Second, we give a collaboration willingness model to characterize developers’ teammate preferences and formulate teammate recommendation as an optimization problem. Then we design a heuristic algorithm to find suitable teammates for a developer. Finally, we have conducted a set of experiments on a Kaggle dataset to evaluate the effectiveness of our approach.
What problem does this paper attempt to address?