Reliability Measurement Of Web Application Based On Probabilistic Model Checking

With the rapid development of Web application technology,the Web application has penetrated into all fields of people's life.Web application failure will not only bring a lot of inconvenience to the people,but also may bring some potential negative impacts,such as the economic losses,decrease of the user experience as well as customer loss and so on.As a result,more and more researchers come to pay close attention to the Web application reliability research.The Web application reliability testing,which is also called Statistical Web Testing,is an effective method to evaluating the reliability of the Web application.At present,Web application reliability testing model is mainly based on Web log data to construct usage model,and then testing and evaluating the reliability according to the usage model.However,most usage model constructing methods did not make full use of the anomaly information in the Web logs.In this paper,we propose a method based on Probabilistic Model Checking to evaluate the reliability of Web application.We apply the anomaly information into the usage model and calculate the reachable probability between states by using Probabilistic Model Checking technology,and then we evaluate the reliability of the Web application according to corresponding reliability model.At last,we analyzed the difference and relationship between two methods.As we know,the reliability of the Web application is not only related to the program itself,but also is associated with the usage of the users.Currently,most classification methods on Web application reliability assessment are dividing users into groups or partitioning based on Web application function modules.Nevertheless,a user may have different usage patterns at different time and scenarios.In view of this point,we adopt Web mining technology to analyze the potential usage patterns in Web applications.We build a usage model for every usage pattern and evaluate the reliability respectively to see the reliability of a Web application under different usage pattern.