Batching techniques for task allocation in workflow systems-towards effective role resolution

Workflow role resolution is the mechanism of assigning tasks to individual workers at runtime according to the role qualification defined in the workflow model. Role resolution decisions directly affect the efficiency of utilizing the most important resources of the organization, i.e., knowledge workers. However, little analytical work has been reported in the literature. In this paper, we model role resolution decisions in workflow systems based on operations research concepts such as assignment and scheduling. We demonstrate that the specific characteristics of workflow role resolution lead to interesting modeling and computational challenges. We propose three online optimization-based policies that utilize batching and study their performance via simulation in comparison with two simpler policies reflecting the current workflow practice. In this simulation study, we investigate how system load, task processing time variation, and the number of workers effect the maximum flowtime of a task, average workload, and workload variation under these five policies. Our study represents an important step towards a comprehensive framework that guides the design of the role resolution module as part of next-generation workflow systems. © 2003 Springer Science+Business Media New York.