Abstract
Knowledge-intensive crowdsourcing (KIC) is expected to provide flexible access to knowledge and expertise. In this research, we proposed a task decomposition method to support the design decisions for KIC tasks decomposition and investigated how the level of granularity affects the crowdsourcing performance. To address the structural complexity, we employed the business process lens and coordination theory to describe the components of KIC tasks and their structural relationships. Afterwards, seven arithmetic tasks with different levels of structural complexity were designed and decomposed into subtasks with different levels of granularity by the proposed decomposition method. A laboratory experiment including 1960 groups of tests to simulate a real crowdsourcing environment was conducted to explore the relationship between different levels of granularity and the crowdsourcing performance. The results suggest that moderate decomposition helps to reduce the completion time and improve the quality of outcomes. A critical point of the level of granularity at which the completion time achieves the minimum is identified.
Recommended Citation
SHI, Xiaojie, "Task Decomposition for Knowledge-intensive Crowdsourcing: Managing Dependency and Structural Complexity" (2021). WHICEB 2021 Proceedings. 6.
https://aisel.aisnet.org/whiceb2021/6