Efficient business process consolidation: combining topic features with structure matching

Abstract

Accurate and effective business process consolidation is an efficient means of overcoming the dynamics and uncertainty in business process modeling. This article presents an approach to automating business process consolidation by applying process topic clustering based on business process libraries, using a graph mining algorithm to extract process patterns, identifying frequent subgraphs under the same process topic, filling the pertinent subgraph information into a table of frequent process subgraphs, and finally merging these frequent subgraphs to obtain merged business processes using a process merging algorithm. Tests on 604 models from the SAP reference model were performed, in which we used the compression ratio to judge the capability of our merging methods; the compression ratios of integrated processes in the same topic cluster were found to be much lower than those of processes related to different topics, and our method was found to achieve compression ratios similar to those reported in previous work.