Efficient Task Decomposition for Sequential Crowdsourced Task Solving

Abstract

In order to facilitate crowdsourcing-based task solving, complex tasks are decomposed into smaller subtasks that can be executed by individual workers. Decomposing task into sequential subtasks attracts a plenty of empirical explorations. The absence of formal studies makes it difficult to provide task requesters with explicit guidelines on task decomposition strategy. We formally present the vertical task decomposition model by specifying the positive quality dependencies among sequential subtasks. Our focus is on addressing solutions of low quality intentionally provided by self-interested workers who are paid equally or based on their contributions. By combining the theoretical analysis on workers’ strategic behaviors and experimental exploration on the efficiency of task decomposition, our study demonstrates the relationship between the incentive and the worker’s performance, and gives the explicit instructions on vertical task decomposition, which show promise on improving the quality of the final outcome.