Planning-based generation of process models

One popular way to model how organisations work is through their internal processes. In the case of current economy, especially in the case of organisations doing business in Internet, the definition of those processes has the problem of continuous adaptation to market needs. Therefore, organisations need powerful tools for automatic modelling, simulating and optimising those processes in such a way that the generated models comply with business rules. This is what the field of Business Process Reengineering (BPR) tries to solve. From an Artificial Intelligence (AI) perspective, planning and scheduling tools can be applied for this task, since they provide a declarative representation of the knowledge on the activities of processes, as well as means to generate only valid process models. However, one of the main obstacles in applying AI planning techniques to real problems is the difficulty to model the domains. Usually, this requires that the people that have developed the planning system carry out the modelling phase since the representation depends very much on a deep knowledge of the internal working of the planning tools. Since, in BPR, there has already been some work on the definition of languages that allow non-experts to enter knowledge on processes into the tools, we propose here the use of one of such languages to enter knowledge on the organisation processes. Once the BPR expert has defined the domain using this language, we use an interface to translate such knowledge for input to an AI planning tool. This planner automatically generates the corresponding process models that are translated back to the process modelling tool.