Mining Workflow Outlier with a Frequency-Based Algorithm

Abstract

Business Process includes workflows which integrated some critical activities among different departments. Workflow management is a quickly evolving technology that can assist business process reengineering and accomplish full or partial automatic processing of a business. Workflow plays a critical role in the ERP (Enterprise Resources Planning) system since it aims to achieve maximum satisfaction for both employee and customers. Any workflow that is irrationally and irregularly designed will not only lead to an ineffective operation of enterprise but also limit the implementation of an effective business strategy. The research proposes an algorithm which makes use of the workflow’s executed frequency, the concept of distance-based outlier detection, empirical rules and Method of Exhaustion to mine three types of workflow outliers, including less-occurring workflow outliers of each process (abnormal workflow of each process), less-occurring workflow outliers of all processes (abnormal workflow of all processes) and never-occurring workflow outliers (redundant workflow). In addition, this research adopts real data to evaluate workflow mining feasibility. In terms of the management, it will assist managers and consultants in (1) controlling exceptions in the process of enterprise auditing, (2) simplifying the business process management by the integration of relevant processes and (3)persisting to improve the quality of business process and enhance enterprise performance.