Dynamic Workflow Management: A Framework for Modeling Workflows

Current workflow management systems do not provide adequate support for workflow modeling. Real life work processes are much richer in variations and more dynamic than is expressed in a typical workflow model. Users need to be able to adjust workloads and modify workflow models on-the-fly. In addition, data about workflow executions are analyzed with process analysis/simulation tools to evaluate design alternatives, so workflow models and data must be structured to reflect the questions that managers and designers are likely to ask. In this paper, we present Dynamic Workflow Management (DWM), a framework for modeling workflows that aims to satisfy these requirements. DWM provides high level packaged task templates for composing task sequences, a MOP-like structure based on dynamic memory theory to organize the task sequences into flexible workflow models, and four perspectives into a workflow model geared toward different users' interests. We illustrate these features of DWM with a case study of an interlibrary loan process and discuss its application in workflow enactment and