Process similarity computing by combining model structure and log behavior

Traditional process similarity measures are not precise enough as they simply compare processes with respect to model structure or behavior separately. To address the problem, in this paper, we introduce a novel measure to quantify process similarity based on model structure and log behavior. The proposed measure first constructs a weighted process graph (WPG) from the process model and the process log, and then computes the weighted node adjacent relation similarity of two WPGs. The WPG that combines both model structure and process log behavior provides a comprehensive description of the underlying business processes for a more accurate similarity measure. By experimental evaluation, we demonstrate the effectiveness of the proposed approach comparing with existing process similarity measures.