Crowdsourcing Team Formation With Worker-Centered Modeling

Area of Science:

• * Computer Science
• * Social Sciences
• * Human-Computer Interaction

Background:

• * Modern crowdsourcing enables complex task solutions through collective labor.
• * Manual team formation is challenging at scale, leading to algorithmic solutions.
• * Algorithmic, top-down team formation often results in poor collaboration and worker dissatisfaction.

Purpose of the Study:

• * To investigate and compare three crowd team formation models: bottom-up, top-down, and hybrid.
• * To evaluate team competitiveness and teamwork quality in simulated open collaboration scenarios.
• * To identify conditions under which different team formation strategies are most effective.

Main Methods:

• * Simulation of an open collaboration scenario, such as a hackathon.
• * Evaluation of team formation models based on worker profiling and task objectives.
• * Analysis of team competitiveness and teamwork quality across different models.

Main Results:

• * The bottom-up team formation model yielded the most competitive teams with superior teamwork quality.
• * Bottom-up approaches are particularly effective for workers with high-risk appetites and high homophily.
• * Algorithmic, top-down methods can lead to alienation, clashes, and dissatisfaction.

Conclusions:

• * Integrating worker agency into algorithm-mediated team formation is crucial for collaborative/competitive settings.
• * Bottom-up team formation enhances worker satisfaction and collaboration effectiveness.
• * Findings have practical implications for designing large-scale crowdsourcing platforms.