CREU Project

• Camille Cobb '12
• Anna Pobletts '12
• Lucy Simko '11
• Faculty Mentor: Sara Sprenkle

Because of the growth of the Internet, users worldwide depend on Web applications for crucial daily tasks. To improve the reliability of Web applications, automated, cost-effective test strategies that adequately test the correctness of Web applications must be developed. One challenge in developing testing techniques is how to automatically generate effective test cases that (1) exercise and expose faults in Web applications and (2) represent users.

A promising approach to generating test cases for Web applications is to generate test cases automatically from user accesses to earlier versions of the application. Millions of users may access a Web application, which could result in generating millions of test cases. Users tend to access a Web application similarly, and, therefore, the test cases generated from these accesses are likely to be redundant. Based on these observations, we propose to build a model from the user accesses and generate a smaller yet equally effective test suite from the probabilistic model. The generated test suite will emulate users' usage patterns closely and will expose faults at less cost than executing all the recorded user accesses as test cases. The ideal model includes (a) a control-flow model of the order that users access a Web application's functionality and (b) a model of the users' data. Previous research has shown that a test case's data significantly impacts its effectiveness, but a general, effective data model has not been developed yet.

Our goal in this project is to design user-access data models applied to automated test-case generation for Web applications.

• Camille Cobb '12
• Anna Pobletts '12
• Lucy Simko '11
• Group Notes

Final Report

We thank our sponsors, the Collaborative Research Experiences for Undergraduates, by the CRA-W and the Coalition to Diversify Computing.