工作流程的概念在許多資訊系統中往往是很重要的議題，工作流程設計不佳將導致企業營運績效不彰，企業策略也無法完全發揮。因此，本研究提出挖掘特例工作流程的演算法，根據過去工作流程的執行頻率，找出特例工作流程作為調整工作流程的參考依據。在管理意涵上能輔助企業主管或顧問： 1.在企業進行內稽內控(Auditing)時，能有效控制特例情形。 2.透過整併相關流程，簡化企業作業流程。例如建置BPM系統時有助 於在流程分析與建構部分的執行，讓BPM系統能將企業流程與資訊系統進行更完美的整合；或是在ERP導入時幫助調整系統參數，關掉不需要執行的流程步驟，使系統調整到最佳化。 本研究之演算法以工作流程的發生頻率，搭配距離為基礎的異常偵測概念，使用經驗法則以及窮舉法方式，挖掘出三種類型的特例工作流程。包括各流程中少發生的特例工作流程、整體流程裡少發生的特例工作流程、以及整體流程中從未執行過的特例工作流程。並透過真實資料，驗證本方法之可行性。 The concept of workflow is very critical in enterprise information system. Irrational workflow will not only leads to an awful operation of enterprise, but also limits the executive of business strategy. This research provides an algorithm which base on the workflow’s executive frequency, to find out company’s workflow outlier for adjusting the whole workflow. It will help the managers and consultants to (1) Control the exception while enterprise auditing. (2) Simplify the business process by integrated related process. The algorithm uses the frequency of workflow, the concept of distance-based outlier detection, empirical rule and method of exhaustion to mine three types of workflow outlier, including less happened workflow outlier of each process (abnormal workflow of each process), less happened workflow outlier of all processes (abnormal workflow of all processes) and never happened workflow outlier (redundant workflow). This research also uses real data to evaluate the feasibility.
