摘要
在圖形化使用者介面(Graphical User Interface，簡稱GUI)的測試腳本中，每個測試案例(Test Case)為一連串對GUI元件的操作與驗證，當使用關鍵字驅動測試(Keyword Driven Test，簡稱KDT)時，每個關鍵字(Keyword)代表一個到多個動作(Action)，而每個測試案例則由多個關鍵字組成。我們知道對於程式碼而言，函式(function)間的呼叫會有耦合(coupling)產生，相似地，KDT測試腳本中測試案例、關鍵字間的呼叫也會有coupling產生，由於設計軟體時，應盡量避免軟體中各模組有高耦合的現象，因此，GUI測試腳本也應該避免高耦合。本論文探討GUI測試腳本的coupling，以及coupling高低對於測試腳本的影響，並提出兩個計算coupling的方法，稱為unweighted coupling與weighted coupling。此外，本論文進行兩個實驗，實驗結果顯示相同測試案例會產生不同coupling的測試腳本，高coupling的測試腳本的可維護性較差，而coupling的高低則對於測試腳本的可讀性沒有太大的影響。
並列摘要
In the test script of a Graphical User Interface (GUI), a test case is a sequence of GUI component’s actions and verifications. In case of Keyword Driven Testing (KDT), a keyword represents one or more actions of the GUI components, and a test case is a sequence of keywords. We know that, for programming languages, when a function calls some other functions, there are couplings between the caller and callee. Similarly, there are couplings when a test case or keyword calls some other keywords in a KDT script. It is known that when designing software applications, high coupling between software modules should be avoid. Therefore, high coupling between keywords and test cases should also be avoided. This thesis studies the couplings in GUI test scripts and the influences of couplings. Two coupling measurement methods, called unweighted coupling and weighted coupling, are proposed. In addition, two experiments are conducted. The experimental results show that two test scripts designed with exactly the same test actions may have different couplings, and the degree of coupling does influences the test scripts’ maintainability. However, the degree of coupling does not influence the test scripts’ readability.