自動化事件激盪流程

Abstract

軟體開發除了撰寫程式碼，亦須理解問題所屬的專業領域。領域驅動設計（Domain-Driven Design, DDD）透過事件激盪（Event Storming）與限界上下文（Bounded Context），將業務邏輯有效融入系統設計。然而，此過程高度依賴開發人員與領域專家的密集溝通，導致成本高昂。為解決此問題，本研究提出一套自動化的事件激盪流程，結合事件因果關係識別（Event Causality Identification）與自然語言處理（NLP）技術，從使用案例規格中自動推導限界上下文邊界，並依據 DDD 模型進行細部設計。本方法可有效降低溝通成本，提升軟體設計流程的效率。此外，本研究亦提出一套評估機制，以衡量自動化產出的限界上下文品質，驗證方法之可行性與實用性。

Understanding the problem domain is as critical to software development as writing code. Domain-Driven Design (DDD) addresses this by integrating business logic into sys tem architecture through methods such as Event Storming and Bounded Contexts. How ever, these practices typically demand extensive collaboration between developers and domain experts, which can be time-consuming and costly. This study presents an auto mated Event Storming approach that combines Event Causality Identification with natural language processing (NLP) techniques. Given use case specifications as input, the sys tem automatically identifies Bounded Context boundaries and generates detailed design models aligned with DDD principles. This approach reduces the need for intensive human collaboration and enhances the efficiency of the design process. Additionally, the study proposes an evaluation framework to assess the quality of the automatically generated Bounded Contexts, demonstrating both the feasibility and practical value of the method.