自動化事件腦力激盪：從 EARS 需求推導出受限制的前後文內容

Abstract

近年來，Event Storming 方法在軟體設計領域逐漸受到關注。然而，Event Storming 仰賴領域專家長時間參與，實施條件嚴苛。為了解決這個問題，我們提出了一種自動化的 Event Storming 流程，可用於從軟體需求中推導出 bounded contexts。該方法先從使用案例規格中提取 domain events，接著運用自然語言處理技術及外部知識庫的領域專業知識，找出領域模型中重要的 entities 及 attributes，進而依循 Domain-Driven Design 的概念將其聚合為 aggregates 和 bounded contexts。該方法有效減少了對領域專家參與的依賴，提高軟體設計的效率。

Event Storming has gained attention as an innovative method for software design. However, conducting Event Storming is challenging due to the need for extensive involvement from domain experts and stakeholders. In response to this challenge, we propose an automated Event Storming process aimed at deriving bounded contexts from software requirements. This process begins by extracting domain events from use case specifications. It then identifies the entities and attributes in the domain model by leveraging Natural Language Processing techniques and integrating domain knowledge sourced from external knowledge bases. These entities are subsequently aggregated into aggregates and bounded contexts in accordance with Domain-Driven Design principles. This method effectively reduces the dependency on domain experts' involvement and enhances the efficiency of software design.