軟體發展流程之多績效指標最佳化方法

Abstract

近年來，由於各國政府大力推廣成熟度整合模式(Capability Maturity Model Integrated, CMM)，有越來越多的軟體組織透過CMMI來管理軟體的發展流程。CMMI共分成五個等級，目前台灣軟體組織僅有非常少數幾家有通過高成熟度等級，此顯示導入高等級之CMMI相當困難，最困難的問題之一是軟體發展流程之績效指標最佳化問題。目前已有文獻建議使用隨機最佳化(Stochastic Optimization Modeling, SOM)的方法來解決上述問題，但此法僅能最佳化單一績效指標，無法同時最佳化多個績效指標。針對多個績效指標最佳化之問題，也有文獻建議以多數決的方法來決定最適之方案與子流程組合，然而，此方法仍需依賴工程師之經驗，且當考量之績效指標越多時，多數決方法越不易選出最佳方案與子流程。本研究應用灰色模糊理論發展出一套可同時最佳化軟體流程之多個績效指標的方法。本論文最後以一個軟體發展流程之實例來說明本研究方法確實有效可行。

In recent years, because software industry promotes the Capability Maturity Model Integrated (CMMI) worldwide, more and more software organizations utilize the CMMI to manage their software development process. There are five grades in the CMMI and only very few Taiwanese software companies achieved high maturity levels (level 4 and 5). One of the difficulties for achieving level 4 or 5 of CMMI is the optimization of software development process. Stochastic optimization method (SOM) is often utilized to optimize the software development process. The SOM can be employed to optimize just one performance-measure. It cannot be utilized to optimize multiple performance measures simultaneously. Therefore, the objective of this study is utilizing the gray-fuzzy method to develop a procedure of simultaneously optimizing multiple performance measures. Finally, a real case from Software Engineering Institute (SEI) is utilized to demonstrate the effectiveness of the proposed procedure.