Process Mining by Measuring Process Block Similarity

Mining, discovering, and integrating process-oriented services has attracted growing attention in the recent year. Workflow precedence graph and workflow block structures are two important factors for comparing and mining processes based on distance similarity measure. Some existing work has done on comparing workflow designs based on their precedence graphs. However, there lacks of standard distance metrics for comparing workflows that contain complex block structures such as parallel OR, parallel AND. In this paper we present a quantitative approach to modeling and capturing the similarity and dissimilarity between different workflow designs, focusing on similarity and dissimilarity between the block structures of different workflow designs. We derive the distance-based similarity measures by analyzing the workflow block structure of the participating workflow processes in four consecutive phases. We first convert each workflow dependency graph into a block tree by using our block detection algorithm. Second, we transform the block tree into a binary tree to provide a normalized reference structure for distance based similarity analysis. Third, we construct a binary branch vector by encoding the binary tree. Finally, we calculate the distance metric between two binary branch vectors. Our initial experience shows that this distance measure can be used as a quantitative and qualitative tool for understanding and detecting block structure similarity and dissimilarity between two workflow designs. It can be effectively combined with a workflow precedence based similarity analysis tool in process mining, process merging, and process clustering, and ultimately it can reduce or minimize the costs involved in design, analysis, and evolution of workflow systems.