Web error classification and usage based model for web reliability improvement

With the prevalence of the Internet, the quality and reliability of web applications have become increasingly important. However, because of the large scale and high complexity, and the relatively short development cycles of web applications, it is more difficult to assure their quality and reliability. Therefore, it poses a challenging obstacle for researchers in related areas to improve the quality and reliability of web applications in an effective and efficient manner. Having observed the Pareto principle in both web error distribution and web usage distribution patterns, this study addresses the web quality and reliability problem from two aspects. (1) A new method for web error classification and analysis is developed by extending an existing defect classification method called Orthogonal Defect Classification (ODC). It utilizes the error related information to identify problem areas in the web, characterize the problems, find common causes, and finally, suggest corrective or preventive activities to improve web reliability. (2) Web usage is modeled by a hierarchical Markov chain model to reflect the relative frequency that different web components are visited. This usage profile is used to guide existing web testing technique and guide resource allocation during web development to achieve overall web reliability. This approach is applied to case studies to show their applicability and effectiveness.