物件導向軟體的耦合力和內聚力之量測

Abstract

耦合力和內聚力冊量測在結構化程式中扮演一個非常重要的角色.資訊 流是耦合力和內聚力的主要來源.而信息傳送則定義了物件導向軟體的資 訊流.此論文中,我們提出一套以資訊流為基的量測方法,專用於物件導向 軟體的耦合力和內聚力.在驗證方面,我們用Weyuker所提出的特性來評估 我們所提出的量測方法.除此之外,我們還利用Basili等人所提出的實驗架 構和已列入IEEE標準,編號1060的驗證方法來設計一個實驗,對目前已提出 並用於測量物件導向軟體之耦合力的方法做驗證,當然包括我們的方法在 內.在驗證過程中我們採用四個準則並搭配相關的統計模式加以分析.而實 驗結果顯示,我們所提出的五個方法中有三個滿足所採用的準則之特性,同 時也較其它之前提出的方法好.這表示我們所提出的這三個量測方法具有 品質評估,品質控制,以及品質預測的功能. The coupling and cohesion metrics played an important role in structured software. Information flow is a major source of coupling or cohesion, while message passings define the information flow in an object-oriented program.In the thesis, we present a set of methods to measure the coupling and cohesionof an object-oriented program based on information flow. An evaluation according to Weyuker's properties is made for the metrics. Besides, following the frameworkproposed by Basili et al. and the validation methodology proposed by Schneidewind,we constructed an experiment to validate our metrics and several coupling metricspreviously proposed in the literature. The quality factors selected in the experiment are mainly concerned with the modification of a class. Four validitycriteria that provide statistical models are used to analyze the data. The experimental results show that our proposed coupling and inheritance coupling metrics are better than the rest. In our five coupling and cohesion metrics forclasses and class hierarchies, three metrics ICH(C), ICP(C) and IH_ICP(C) satisfyall the validity criteria; that is, they provide the quality assessment functions, quality control functions and quality prediction functions.Keywords: coupling and cohesion metrics, the validation methodology, validity criteria, quality assessment functions, quality control functions, quality prediction functions.