Procedure Of Distance Measurement Methods

Process alignment has been a focus of research in the field of workflow, and is also a hot topic of business process management in recent years. It means that the similarity or dissimilarity of the two processes can be concluded through comparing certain characteristics between two processes, which is of great significance for the analysis and improvement of existing processes.Process distance is a description about the extent of the difference between two processes. Previous studies have started only from a single characteristic. They haven’t considered a number of characteristics and correlation between them, so there are some limitations. In fact, the workflow can be seen as a set of activities in accordance with some kind of order. Consequently, on the basis of previous studies, this paper carries out in-depth analysis for multiple processes with the same background and the same tasks after considering control flows, activities and associations between them. Therefore, a clear measurement method of process distance is put forward and the calculation formula for the distance process is further deduced. In addition, it does a certain degree of research about process clustering based on them.(1) There is a detailed description of the measurement methods of process distance. A qualitative analysis is introduced to verity the effectiveness of the process distance and a quantitative analysis to further study the distance. So the distance between the process X and Y is as follows:d(X,Y)=(?)dc(X,Y)2+da(X, Y)2.①In the quantitative analysis of process distance, the process distance is decomposed into two components:control flow component(namely dc) and activity component(namely da).Through the decomposition on workflow nets by structured and hierarchical idea, a process tree is established. Process tree can well integrate the dc and da (where dc, da are respectively studied based on the non-leaf node, leaf node).②Through the analysis of the context of the activities, the activity context is defined: ActivityContext={(R, D, T, P),(Data, Operation)}. By this definition, the same name of the different activities and the different activities of the same name can be distinguished, which achieves standardization to name activities.③A custom operator approach is used to calculate dc and da separately. By user-defined, the exact value about dc, da can be calculated. (2) The study of the process distance is extended. Let distance matrix be the input, thus improving k-medoids algorithm, a clear way of process clustering is put forward.(3) The process distance is applied into specific case. The distances between multiple processes can be calculated through the measurement method of process distance, then all processes can be clustered, which proves effectiveness of the measurement method and the feasibility of clustering methods proposed.