Dynamic Workflow Net-based Workflow Process Mining

The workflow life cycle consists of four phases: workflow design, workflow configuration, workflow enactment and workflow diagnosis. In the traditional approach the design phase is used for constructing a workflow model. This is typically done by a business consultant and is driven by ideas of management on improving the business processes at hand. Creating a workflow design is a complicated t ime-consuming p rocess and t ypically t here are d iscrepancies b etween the actual workflow processes and the processes as perceived by the management.The workflow mining technology is not just a tool of workflow design, but it is very useful for understanding the current business processes. The goal of workflow mining is to reverse the process and collect data at runtime to support workflow design and analysis. Since the information construction has been carried on in many organizations for several years, the organizations have their own information systems, such as MIS, ERP or CRM. In most information systems transactional data is registered. The information collected at run-time can be used to derive a model explaining the events recorded.The beginning of workflow mining is to collect and format the workflow log. We defined the workflow log as a set of task series. That definition contains the necessary information of the log with a simple format. We also defined the related conceptions.It is a challenge to exploit more information in a set of workflow log. The former workflow m ining m ethods t ypically t ake t he w orkflow a s a w hole. B ut i n p ractice many business processes can vary with the change of time, environment.In this paper we take a different perspective with respect to the problems related to the dynamic information in workflow log. Dynamic workflow net(DWF-net) is proposed based on the conception of WF-net. DWF-net consists of dynamic transactions, dynamic place and directed arcs between them. The dynamic transactionhas two attributes: (s, d). s means the frequency of the dynamic transaction present to the workflow log. It is also the support factor. When s satisfies the allocated condition, the dynamic transaction is regarded as "real". D describes the variety of s and is called "dynamic description" of s. The dynamic place also has two attributes: (c, d). c is believable factor and d is "dynamic description" of s. In this paper the definition of the routes and the proof of the soundness of DWF-net is put forward.We designed the algorithm of generating a DWF-net from workflow log. The first step is to get the structure of the DWF-net. The second is to get the dynamic descriptions. In this paper the idea and the details of algorithm is both introduced, the complexity is analyzed as well.DWFgenerator is a prototype, which realizes the DWF-net generateon algorithm. The system is mainly developed on Maltlab6.5. It can mine the workflow log, which satisfies the format proposed in this paper and gets the DWF-net.