一個時序工作流程編輯時異常動作的遞增性分析

Abstract

一個好的結構化工作流程是由更小的結構化工作流程和程序所組成，結構化工作流程裡的控制結構是由平行與選擇結構組成。一個工作流程可以被轉換成一個結構化工作流程，用來分析artifact的異常。此外，時間因素也被加入到每個程序中，以提高偵測artifact異常的準確度。在這篇論文中，我們首先引進了將一個時序工作流程轉換成一個時序結構化工作流程的方法，然後在此時序化工作流程上找出異常。這樣的工作流程被稱之為TS工作流程，以及我們也發展出一個批次的演算法計算異常。由於CASE工具的盛行，在工作流程上的遞增性分析也變得更為重要。然而，在一個時序工作流程裡作遞增式分析並不容易，因為迴圈及時間的因素使得分析變得更為複雜。我們對編輯工作流程的動作分類並加以討論，發展出一系列的演算法，這些演算法是作用在被編輯後的TS工作流程上，用來分析有何artifact異常產生。

A well-structured workflow is composed of a group of well-structured workflow(s) and process(es), where the control structure is composed of parallel/decision structure. A general workflow can be transformed into a well-structured workflow for the analysis of abnormal artifact behavior. Besides, the temporal factors can be added into each process to help improve the preciseness of anomaly detection. In this thesis, we first introduce how to transform a general workflow with temporal factors into a well-structured one with temporal factors for anomaly detection. The workflow is called a TS workflow, and a batched analysis algorithm is developed. Due to the popularity of CASE tools, the incremental analysis for the workflow is getting important. However, the incremental anomaly detections working in a workflow with temporal factor are difficult, since the loop and related temporal factors introduce complicated problems. We discuss and categorize the edit activities and develop a series of analysis algorithms which are organized in to perform the anomaly detections in the corresponding TS workflow maintained after each edit activity.