abstract

As web applications become increasingly complex, automated testing becomes difficult due to the overload of information. The state of the art mechanisms of user-based testing involve automatically generating test suites in two main steps. First, the tester generates a sequence representing a user’s path through the application, and later she fills in parameter values. We focus on the second stage of generation, called the data model, and we propose a number of characteristics to classify parameters. Parameters are categorized based on their use in the applications as well as their treatment by the user. Based on the classification of the parameter, we can then pick the best combination of factors to determine the best or most realistic parameter value. Our factors are comprised of the other possible elements of the user’s path through application that may affect the value of the parameter. By determining which of these factors is most relevant given the type of parameter, we are able to build a more accurate data model and thus more accurately model our dersired test suites.