Efficient scheduling in crowdsourcing based on workers' mood

Crowdsourcing workers need to properly schedule when to work and when to rest in order maintain good productivity and ensure the long-term operation of crowdsourcing systems. However, existing algorithmic crowdsourcing works largely overlook this problem and only focus on strategies to optimize the distribution of tasks among workers. In this paper, we propose the Affective Crowdsourcing (AC) approach to fill this important gap. It is a distributed dynamic scheduling approach which minimizes crowdsourcing worker effort expenditure while achieving high collective productivity. Our approach leverages the emerging body of evidence about the relationship between people's mood and their productivity to operationalize the `productive laziness' concept from workforce management research. Extensive experimental studies based on a large-scale dataset released by Tianchi demonstrate that AC significantly conserves worker effort, while making the smallest sacrifice in terms of task completion rates as compared to other alternative scheduling approaches. The proposed approach is novel and establishes a framework for crowdsourcing workers to optimize their work and rest schedules.