Workflow Application Control Flow Classification Based On The Distance Metric

The workflow classification is a technology which uses the specific method of data mining according to workflow’s structure feature and node semanteme to classify workflow into different categories.The workflow in the same category has a similar structure,which provides convenience for process retrieval and process reuse.Firstly,the essay do a simple introduction for Workflow management model and classification. Then it introduces the process that separates control node from the workflow and generates control flow diagram. Secondly, based on the distance between the control node, it builds a model measuring the distance of the workflow’s control flow through control flow diagram, and proves distance-measuring model meeting the nature of the reflexivity, symmetry and triangle inequality in theory. the essay shows the distance calculations process and its effectiveness in the application of the process matching. Finally,applying the measuring the distance of the workflow’s control flow to classification combining with kNN(k-Nearest Neighbors) method,and having checkouted the correctness from the result.Introducing a kind of classification which is based on distance-KNN(k-Nearest Neighbors),and showing the significance of workflow classification in process reuse.It applies the kNN to the classification combining with method measuring the distance of the workflow’s control flow.Finally,it can be verified that the method in this essay is valid though the classification result.We accomplished a prototype system which is based on workflow’s control flow distance and uses the kNN method to classify workflow.All of the workflow are represented by XML document which is based on JBPM standard.The system parses XML document and separates control flow from the workflow,then calculates distance between two control flows,and uses the kNN method to classify the workflows.