運用一具Hierarchy 特性的Timed CPNets 技術來分析BPMN 工作流程

Abstract

相對於BPMN而言，現存的商業流程之研究或商業軟體大多只提供或使用其中的部分。BPMN主要的組成元素包括:控制流程、訊息流程、資料流程以及角色分配。它也提供多實體activity、事件觸發activity及進階控制機制。雖然這些元素讓BPMN具更大的流程表達能力，但也增加了設計階段其所表達之流程的分析困難度。本論文提出一個正規流程模型來協助根據BPMN四種組成元素所描述的商業流程。同時，也提供一具階層特質之時間顏色派翠網模型。並建立一套流程與此網的轉換規則，以便將上述BPMN商業流程轉換成相對之時間顏色派翠網，來運用既有之分析方法做靜態分析—如deadlock檢查。在本論文中，我們更進一步探討well-formed和unstructured相當普遍的流程之分析。此外，我們將以一個實際的例子做示範，利用時間顏色派翠網deadlock分析方法，再根據其結果推斷可能會影響流程執行的異常artifact之使用。最後，我們也將比較刻下技術與我們之研究成果。

Although many business process models have been proposed, most of them do not apply all the following arguments: control, message and data flows and role assignments, defined completely in BPMN. Besides, they do not provide the multi-instance activity, event-triggered activity or the control node with complex mechanisms as in BPMN. On the other hand, these features allow a process to be defined with richer semantics but increase the difficulty of correcting an error or inaccurate process in workflow design. This thesis proposes a formal process model to help describing a BPMN-based process. To simplify the analysis, we also provided Hierarchical Timed Coloured Petri Nets ( PNets), which is extended from Time Coloured Petri Nets with hierarchy and allows some analysis with existing techniques. Once a workflow based on our BPMN model is specified, a series of mapping rules can be used to transform the workflow into a PNets for analysis. An example is applied to demonstrate the transformation and the corresponding deadlock detection. Furthermore, the artifact usage anomaly detection mechanisms within either a well-formed or unstructured process are discussed. Finally, a comparison among related works and ours and the future works are presented.