LAUSR

Crowdsourcing has become a new distributed paradigm, which uses online crowds to solve complex problems. Recently, in order to reduce the development workload and research threshold of crowdsourcing applications, crowdsourcing process modeling is attracting more and more attention. However, complex crowdsourcing processes used for creative and open-ended work have remained out of reach for process modeling, because this type of process usually has a dynamic execution, in which the type, number, and order of tasks and subtasks are often unknown in advance but are determined dynamically at runtime. In this paper, we propose a modeling approach and supporting framework to fill this gap. Specifically, we provide a task model composition to allow task creation on demand, while collaborating on tasks in a tree structure to adapt to the dynamic execution. Moreover, we introduce a set of message communication modes to support data exchange among tasks. Finally, we construct a framework named CrowdModeller to embody this approach. Through two evaluations, we demonstrate its effectiveness.