Grant: Discovery Grants Program - Individual

A workflow is a collection of activities performed by individuals, organizations or automated devices, such as computers, to accomplish a clearly defined goal. Workflow management systems have been developed to manage these activities, but they suffer from serious deficiencies if the activities are part of time-sensitive, data-aware collaborative processes. These problems are particularly acute for processes that are safety critical, i.e., processes where human lives are at stake. We seek to develop techniques to model and reason about collaborative workflows, which interface with dynamically changing data at predefined decision points. We use graphical modelling languages, which can be automatically translated to executable code, reducing the possibility of coding error. The long-term goal is executable software systems which provide measures of quality assurance. We will develop methods to check that the workflow model actually achieves precise goals or specifications, through the use of a technique called model checking. Model checking time sensitive workflows is usually very complex, and we must find ways to reduce the complexity. One way is to develop runtime monitoring methods which inspect the system at specific points to determine if at that point a problem can develop. To be adaptive, and to facilitate collaboration, we organize facts and rules about the data in a knowledge representation structure called an ontology. An ontology permits the calculation of knowledge from the explicit facts; this reasoning can take a long time: we need to develop fast methods of reasoning, so the implicit knowledge is available in time to guide the workflow. The motivating applications for this research are workflows for healthcare processes, which occur in a variety of settings, must be accomplished in specified intervals of time, and must adapt to patient data and other frequently changing information; lessons learned can be applied to many safety critical processes.