Supporting Effective Unexpected Exception Handling in Workflow Management Systems Within Organizational Contexts

• Main Author: Mourão, Hernâni
• Publication Date: 2008
• Language: por
• Source: Repositórios Científicos de Acesso Aberto de Portugal (RCAAP)
• Download full: http://hdl.handle.net/10451/14289
• topic: workflow management systems; unexpected exceptions; collaboration support; unstructured organizational activities

Workflow Management Systems support the execution of organizational processes within organizations. Processes are modelled using high level languages specifying the sequence of tasks the organization has to perform. However, organizational processes do not have always a smooth flow conforming to any possible designed model and exceptions to the rule happen often. Organizations require flexibility to react to situations not predicted in the model. The required flexibility should be complemented with robustness to guarantee system reliability even in extreme situations. In our work, we have introduced the concept of WfMS resilience that comprises these two facets: robustness and flexibility. The main objective of our work is to increase resilience in Workflow Management Systems. From the events demanding for Workflow Management Systems resilience, we focused on ad hoc effective unexpected exceptions as those for which no previous knowledge exist is the organization to derive the handling procedure and no plan can be a priori established. These exceptions usually require human intervention and problem solving activities, since the concrete situation may not be entirely understood before humans start reacting to the event. After discussing existing approaches to increase WfMS resilience, we have identified five levels of conformity. The fifth level, being the most demanding one, requires unrestricted humanistic interventions to workflow execution. In this thesis, we propose a system to support unrestricted users' interventions to the WfMS and we characterize the interventions as unstructured activities. The system has two modes of operation: it usually works under model control and changes to unstructured activities support when an exception is detected. The exception handling activities are carried out until the system is placed back into a coherent mode, where work may proceed under model execution control