DRECT: A search-based developer recommendation approach for software crowdsourcing platforms

Collaboration efficiency in modern software development is increasingly relying on distributed individuals and their collaborations. Identifying suitable developers for given tasks poses significant challenges, particularly due to the complexity of selecting the right candidates from a large pool of potential contributors. Moreover, the abundance of tasks on crowdsourcing platforms makes the process of finding appropriate developers time-consuming, further complicating the capture of developers’ expertise and the timely completion of tasks. While several approaches have been proposed to automatically recommend developers for tasks on these platforms, our study presents a complementary approach that has the potential to enhance recommendation performance. In this paper, we introduce DRECT, our approach designed to automatically recommend developers for specific tasks. We frame this as a multi-objective combinatorial problem and employ the NSGA-II (non-dominated sorting genetic algorithm) as the search method to identify the optimal set of developers. Our approach optimizes three key objectives: (1) Maximizing developers’ expertise and collaboration, (2) maximizing the similarity between developers and tasks, as well as task-to-task similarity, and (3) minimizing developer workload by taking into account the number of active tasks they are currently handling. To assess the effectiveness of DRECT, we conducted an empirical study using a large, real-world benchmark dataset from Topcoder. The results demonstrate that DRECT significantly outperforms several popular search-based algorithms and recent state-of-the-art approaches, highlighting its potential as a robust solution for developer recommendation on crowdsourcing platforms. These findings highlight the importance of our work by providing essential guidelines for researchers, contributors, and maintainers to enhance the developer recommendation process.