Software fault predictors for web applications

Abstract

Our daily life increasingly relies on Web applications. Web applications provide us with a seemingly unlimited amount of information and abundant services to support our everyday activities. As the Web usage continues to grow, users expect that Web applications will be more mature and useful. Therefore, quality assurance for Web applications is becoming important and has gained much attention from software engineering community. In recent years, in order to enhance software quality, many software metrics have been proposed to be used as fault predictors in software fault prediction models. Such models can be applied to predict which software modules are likely to be faulty during operations. If we can predict fault-prone modules in software systems, we can raise the effectiveness of software testing activities and reduce project risks. In addition, software project managers can deliver software projects within budget with minimal schedule slippage. Although Web applications are now prevalent, so far no fault predictors are proposed for them with consideration of their particular characteristics. In this study, we try to introduce new software metrics which can be used as fault predictors for Web applications after analyzing their characteristics. The experimental analyses show that using a combination of our proposed fault predictors and popular fault predictors achieves better results than using only popular ones.