Free Trial

Safari Books Online is a digital library providing on-demand subscription access to thousands of learning resources.

Predicting Temporal Exceptions in Concurrent Workflows INTRODUCTION In a highly dynamic business environment, simultaneously executing departmental work- flows form digital business ecosystems (Boley & Chang, 2007) which are designed to share limited resources. In this chapter we address the temporal exception prediction problems of a digital busi- ness ecosystem involving concurrent workflows. A workflow specification defines how a busi- ness process functions within an organization (Son, Kim, & Kim, 2005). Based on these specifications, Workflow Management Systems (WfMS) allocate and dispatch work to users (Li & Yang, 2005). A workflow instance is an execution of a workflow specification. During the execution of a workflow instance, some events that are not defined in the workflow specification may occur. These events are typically considered as exceptions in workflow management systems. fleet of vehicles. Delays occur when only a limited number of resources are available during a given interval. We denote deadline violations caused by such delays as temporal exceptions. In workflow management systems, control- flow patterns are used to describe the order of tasks that make up a process and the relation- ship between them (Van der Aalst, Ter Hofstede, Kiepuszewski, & Barros, 2003). In this chapter, we focus on predicting temporal exceptions for workflows which include iteration patterns (loops). Iteration pattern refers to a repeated ex- ecution of one or more tasks within a workflow. Although iteration patterns are extensively used in workflow specifications, less attention has been devoted to understanding their implications for temporal exceptions. Specifically, rapid changes in the number of iterations can cause deadline violations as well as conflicts in resource usage. Our approach can be divided into two phases: