Software systems, especially web applications, play a major role in our personal and professional lives.
and professional life. It is essential to minimize failures, but at the same time, they are becoming more and more complex and therefore difficult to test. Exploratory testing has proven to be an effective method to find bugs that require complex interactions with the system. This testing practice relies on the business knowledge and experience of testers to validate the system rather than on scripted test suites. However, testers' tooling is underdeveloped in this area. We argue that supporting exploratory testing to inform the tester can improve test quality and decrease the level of expertise required for testers. Knowledge of the business helps testers identify areas of interest in the system, and testers' experience helps them maintain diversity in their testing. However, testers can be biased during their testing, for example because they are not familiar with certain parts of the system, or with the behavior of users on the application behavior of users on the application. In addition, it is common for exploratory testing sessions to involve several involve several testers. It is then difficult to avoid some redundancy in the tests. Moreover, a certain knowledge of the system is required to identify relevant tests, so it is complicated to use external testers, as with a crowtesting platform.
In this thesis, we propose and evaluate approaches to help testers perform effective exploratory testing. We propose to model the interactions performed by exploratory testers, recommend actions of interest, and provide direct feedback during the session. We base our recommendations on Markov models built from sequences of interactions with the web application, and on state machines. To evaluate our approaches, we have developed a software base, which will serve as a basis for future work based on web application testing. Our results show that the proposed approaches help testers to explore on real web applications. Our goal for future work is to improve the proposed guidance, bringing more automation in exploratory testing.