Study On Mining Algorithms Of Batch Processing Patterns For Workflow Instances

A workflow is a partial or total automation of a business process, in which a collection of activities is a logical unit or step of work. Batch processing in workflow is the accomplishment of several requirements in one action by combining multiple workflow activity cases together, while in traditional approaches, they are handled by several actions. Batch processing of activity cases based on certain rules can usually economize system resources and improve the system efficiency, but it is almost neglected by current workflow and very little support is offered in current workflow systems.Exploratory research on the workflow management systems supporting dynamic batch processing (DBP) and different DBP patterns in workflow has been done and it has drawn many researcher's attention. However, there are still some problems need to be investigated. The following three problems are not solved yet: 1) determination of which activity deserves batch processing (hereafter batch-deserving activity); 2) if it is a batch-deserving activity, how the multiple cases of the activity are batch-processed; and 3) the setting of the batch processing areas (batch processing patterns). In fact, just like the definition of workflow models, the modeling of batch-deserving activity as well as batch processing area is also a time-consuming and error-prone task. They can be easily influenced by the perception of business process designer (hereafter designer). Moreover, designers may not know exactly which activity deserves batch processing at workflow building time since there is no real data at that time to give us confidence. Even more designers may ignore these kinds of batch processing features due to certain reasons and this happens very often. To optimize business processes, therefore, it is badly important to explore a way to identify and model both batch-deserving activities and batch processing areas automatically.Aiming at solving the above problems, this paper proposes and explores the problem of mining batch processing patterns from workflow logs. This paper has carried on a research into it from the following two aspects. First, in current workflow environment, does research on the identification of batch-deserving activities and their batch processing features as well as recognition of batch processing areas in workflow. Second, in workflow system supporting workflow activities' batch processing, does research on the mining of workflow models with batch processing activities and its responding sub-process.The main contents and contribution of this paper includes: 1) describe the batch-deserving activity in workflow and define its batch processing features; 2) propose a algorithm for identification of activity batch processing features and a algorithm for recognition of batch processing areas in workflow in current workflow environments; 3) make a definition of batch processing pattern in workflow system supporting workflow activities' batch processing and propose a responding algorithm for mining workflow activity cases' batch processing pattern; 4) several simulative experiments are done to verify algorithms' effectiveness. The experimental result shows these algorithms can solve the problems of activity cases' batch processing in different workflow environments and facilitate the application of workflow system supporting workflow activities' batch processing.