智慧型系統需求工程Requirements Engineering

智慧型系統（如 AI 驅動的系統）的非功能性需求（NFR，如安全性、準確性、可靠性與倫理性）往往抽象、難以量化，且彼此牽動、充滿不確定性，傳統需求工程不易完整捕捉。本研究以智慧家庭為主要應用情境，探討如何運用知識工程與本體論（Ontology）的塑模技術，建立一套標準化流程來定義、追蹤並驗證這些關鍵需求，藉此提升智慧系統需求描述的精確度、一致性與可信度。Non-functional requirements (NFRs — safety, accuracy, reliability, ethics) for intelligent (AI-driven) systems are abstract, hard to quantify, interdependent, and highly uncertain, so conventional requirements engineering struggles to capture them fully. Taking the smart home as the primary application scenario, this research studies how knowledge engineering and ontology modeling can establish a standardized process to define, trace, and verify these critical requirements — improving the precision, consistency, and trustworthiness of requirement specifications for intelligent systems.

• 本體論塑模：建立智慧型系統 NFRs 的知識表示模型。Ontology Modeling: Establishing knowledge representation models for intelligent system NFRs.
• 需求追蹤：將抽象的非功能性需求映射至具體的系統規格。Requirement Traceability: Mapping abstract NFRs to concrete system specifications.
• 衝突分析：識別並解決需求間的衝突（例如效能與安全性間的權衡）。Conflict Analysis: Identifying and resolving trade-offs between requirements (e.g., performance vs. security).

我於碩一在此國科會計畫擔任研究助理，參與非功能性需求的塑模研究與本體論建構。這段經歷讓我從軟體工程的上游——需求定義——理解一套「可信賴的智慧系統」該如何被嚴謹地描述與驗證，與我後續在 AI 與系統開發的實作經驗互補。As a first-year M.S. research assistant on this NSTC project, I contributed to NFR modeling research and ontology construction. The work gave me a view from the upstream of software engineering — requirement definition — of how a trustworthy intelligent system should be rigorously specified and verified, complementing my later hands-on work in AI and systems development.