從多媒體的視覺化需求描述轉譯為UML設計描述之轉譯器設計及實作

Abstract

在軟體開發過程中，誘導及確認使用者需求是一重要的階段。許多研究指出[1-3]，以視覺化的方式，不但可以即早獲得使用者的需求回饋以及快速地確認使用者的需求；同時可避免因閱讀大量的文字需求描述而產生的誤解。確定使用者需求後，緊接著進入分析、設計階段。目前以UML[5]作為分析、設計的工具，已成為一種趨勢。然而，可能因不正確的需求、或因知識領域的不同、或因開發者本身的經驗，而導致必須一再地修正使用者案例圖(Use Case Diagram)，進而連帶影響整個軟體開發流程。 有鑑於此，本論文引用多媒體視覺化需求描述建構系統[1]來即早搜集、確定使用者需求。並著重於研究將多媒體的視覺化需求描述轉譯為UML設計描述之方法，以改善軟體開發流程。在此研究中細分幾個重要步驟，如轉譯規則的建立，系統模型設計等，也將在此研究一一地加以闡述。 為了驗證及展示此研究的可行性，我們實作一個轉譯器，將視覺化需求描述轉譯為UML設計描述；並以XML Metadata Interchange(XMI)格式[6]來貯存，使之得以UML CASE Tool來呈現UML Diagrams。此轉譯器亦從XMI資料產生相關程式碼。使用UML CASE Tool產生程式碼框架後，可參考相關程式碼來建構target system。如此，進而可改善軟體開發流程。

In the software system development, eliciting the requirements from user is an important phase. It had been shown that using visual approach to represent user requirements not only can receive user’s early feedback for obtaining user’s correct requirements, but also can avoid reading large amount of text-based requirements that usually lead to misunderstand on requirements between user and developer [1-2]. After requirement discovery phase, software developers will start analyzing and designing the target software system. Using Unified Modeling Language (UML) approach [5] to analyze and design the target system become more and more popular. However, due to the incorrect user requirements, domain knowledge issues, or limitation on software developer’s experience, Revision of requirement become unavoidable. Eventually, the whole software development process will be affected. In view of that, this thesis quotes the visual requirement authoring system (VRAS) [1] to resolve above-mentioned issues such that we can collect and confirm user requirement as early as possible. This thesis research focuses on the translation of multimedia visual requirement representation into UML representation such that the software development process can be improved. There are important issues (such as translating rules building and proposed system model presenting) will be elaborated in this thesis. To demonstrate the applicability of the proposed model, a translation system is designed and implemented for translating multimedia visual requirements representation into UML representation. The target code is saved as XML Metadata Interchange (XMI) format [6], which can be used to represent UML notations by UML CASE Tool. Through this translation system, we generate the UML design representation automatically. Developer can use UML CASE tool to generate code framework to reconstruct the target system. Eventually, we can improve software development process.