本研究將流程探勘(process mining)領域中利用不同演算法所挖掘得到的模型做量化上的比較,採用樹狀圖(Block tree、Binary branch tree)比較法,並以穩態機率的機率值視成權重值,不但解決過去特定圖形權重值不易計算的問題,同時亦提供了一個辨識方法,可以辨識活動與活動間是否存在迴圈或其他類似迴圈具相依事件的模型。研究結果顯示本研究提出的方法的確可以客觀且有效的比較相同資料下,不同演算法(α-algorithm、α++-algorithm)所得到模型之間的差異。最後,本文也將提出的比較方法建立一個簡易的流程倉儲系統,縮短尋找合適模式的時間。冀望將來面臨有關流程分析或建置上的問題時,可以最短的時間,創造最好的分析效益。
In this research, we focus on the comparisons of the quantified models from several different algorithms in the process mining. We adopt the concept of Block tree and Binary Branch tree and the steady state probability to establish the graph models and the computation of activity weight values. In the proposed method, we can not only solve the difficulty to calculate the activity weight value for some cases, but also provide a way to identify a dependent activity such as loop or similar activity in the process. Our study shows that our proposed method can provide a more objective and effective on identifying the discrepancy between models from different mining algorithms(α-algorithm、α++-algorithm) based on the same dataset . We also build a simple process warehouse system in order to shorten the searching time of a suitable model in the research. We hope that our system can create the best benefits with shortest time when we face solving problems in the process analysis or constructions in the future.