A Developer Recommendation Framework in Software Crowdsourcing Development

Crowdsourcing software development (CSD) makes use of geographically distributed developers to contribute for massive tasks and thus brings about flexibility, convenience and efficiency for both task requesters and software developers, and its competitiveness for requesters’ adoption guarantees the quality of software effectively. Many CSD platforms, however, just play a role of intermediate, so requesters using these platforms need to go through all available developers to choose the appropriate ones, which makes less efficiency and risks the lack of experienced participations. In this work, we present a feature model to depict software crowdsourcing tasks and accordingly propose a recommendation framework to recommend developers in CSD by combining a neural network and a content-based method. In the end of this work, we test our approach on TopCoder’s historical dataset for recent 3 years and the results show that our approach increases the accuracy more than two times besides having a pretty good extendibility.