在軟體需求與日俱增的情形下，軟體的規模由小變大，系統複雜度由簡單變複雜，面對這種需求的發展，軟體設計構思不再單純只是其開發程度的完全與否，品質的議題已開始受到重視。在軟體品質的控管上，首先有軟體度量(Measurement)的思維出現，透過度量的方式，取得的資料間接反映出專案在開發中或者是維護中相關的重要訊息，可想而知，以管理的層面而言，可根據客觀的度量結果而做出合適的決策。 度量規劃運用目標-問題-度量指標方法論(Goal/Question/Metric, GQM)此方法來輔助進行有一定的幫助，因為度量指標相當繁多，要「指出專案所欲獲得的資訊」並不容易，接下來的推論度量指標自然也就更為困難，所以，如果將GQM的方法運用到度量流程的規劃階段，便能有方向性的去選擇需要的度量指標。 在本論文中，我們將提出一個以Fuzzy GQM為基礎之軟體度量規劃系統(Fuzzy GQM Based Software Measurement System)。雖然GQM在理論上能有效的輔助度量規劃人員進行規劃，但其實仍有其應用上的問題。倘若度量規劃人員的度量經驗不足夠，且對軟體開發的程序並不十分了解，即使GQM提供了一種選擇度量指標的方法，仍然無法解決經驗不足的問題。本系統便是一套將GQM理論在應用上的問題所改良之度量規劃輔助系統，利用知識庫紀錄過往度量規劃的專案資料，運用模糊理論的概念定義目標、問題以及度量指標之間的關係，並利用Fuzzy Petri Net來儲存以及表現GQM關係，將目標、問題以及度量指標以動態的方式供度量規劃人員選擇，以此協助度量規劃經驗不足的規劃人員進行度量規劃。Fuzzy Petri Net已被廣泛的運用在許多的知識庫系統，他能夠以結構化的方式表達知識以及Fuzzy推論規則。 The continuing improvement of software quality is extremely important for software developments. Measures are the fault detectors that let us know when our projects and processes encounter problems. Further, measures can give us an objective indication whether an improvement actions working or not. To accomplish quality in software processes and products, it is required to develop systematic measurement programs, congenial with the organizational purpose and property and tailored to the quality aspects that are being advised. The Goal-Question-Metric (GQM) approach was proposed by Victor Basili et. al.. The main purpose of the GQM is to guide the definition and development of a goal-driven measurement program. Although the GQM seems to be simple to apply, it involves a sequence of well defined, relative stages, phases and activities that are not easy to understand and apply for non-expert. In this paper, we proposed a Fuzzy GQM Based Software Measurement System. This system applies fuzzy Petri net to represent the fuzzy GQM relation. Exploiting such mechanism, our system can dynamically provide goal list, question list and metric list for users depend on specific project features. Users only need to choose from options of goals and questions, a list of metrics will be recommended by the decision support system. Moreover, this system enables to capture experiences in the knowledge base so that such experiences could be reused in other projects. The Fuzzy Petri-Net is adopted for the analysis the relationship between goal, question, and metric in this system.
