Developing a process pattern mining approach to support workflow structure analysis

• Main Authors: Ke-chiuan Lai, 賴科全
• Other Authors: none
• Format: Others
• Language: zh-TW
• Published: 2009
• Online Access: http://ndltd.ncl.edu.tw/handle/43197599841913479796

碩士 === 國立臺灣科技大學 === 工業管理系 === 97 === The complexity of business model has been greatly increased since there are numerous interactions among businesses than before. How to analyze business models exactly and fast will be an important issue. In order to realize activities of business, users could observe the business process of operation by Petri Nets models. However, the activities of business in Petri Nets models usually just be represented by diagrams. The complexity of Petri Nets model cannot be identified and the soundness of Petri Nets is hard to be analyzed. Based on above factors, this research has proposed an approach to abstract the patterns of process models based on Petri Nets. The approach of mining the workflow patterns in this paper contains five patterns. First of all the input and output structures of five patterns are defined by a matrix-based representation. Then, a matrix-based assign-match algorithm using a matrix-based representation of five patterns is constructed and used to discover patterns of workflow. Furthermore, the purpose of this research provides an approach of analyzing the structure correctness of a process by using the result of mining the workflow patterns. Thus incorrect structures of a process will be distilled. By using the flow relation of workflow patterns to verify the business processes structural conflict will help to restructure or redesign processes.