Improving Parallel Pattern Discovery from Directly Follows Graph Model

Process model discovery is an important part of process mining. State-of-the-art algorithms still have trouble detecting long parallel patterns with incomplete concurrency patterns. This paper proposes a method to exploit the characteristics of parallel block patterns to discover an accurate parallel block. The experiment compared the proposed method with state-of-the-art algorithms, Inductive Miner and Split Miner. The quality of the discovery results is measured using fitness, precision, and f-score. The experimental results show that the proposed method outperforms the state-of-the-art algorithm in discovering long parallel block patterns with incomplete concurrency patterns.