Applications with a Graphical User Interface (GUI) front-end are ubiquitous nowadays.

While automated model-based approaches have been shown to be effective in testing of such applications, most existing techniques produce many infeasible event sequences used as GUI test cases. This happens primarily because the behavioral specifications of the GUI under test are ignored. In this dissertation we present an automated framework that reveals an important set of state-based constraints among GUI events based on infeasible (i.e., unexecutable or partially executable) test cases of a GUI test suite. GUIDiVa, an iterative algorithm at the core of our framework, enumerates all possible constraint violations as potential reasons for test case failure, on the failed event of an infeasible test case. It then selects and adds the most promising constraints of each iteration to a final set based on the Validity Weight of constraints. The results of empirical studies on both seeded and nine non-trivial open-source study subjects show that our framework is capable of capturing important aspects of GUI behavior in the form of state-based event constraints, while considerably reducing the number of insfeasible test cases. The second part of this dissertation deals with the problem of automatic generation of relevant test data for parameterized GUI events (i.e., events associated with widgets that accept user inputs such as textboxes and textareas). Current techniques either manipulate the source code of the application under test (AUT) to generate the test data, or blindly use a set of random string values. We propose a novel way to generate the test data by exploiting the information provided in the GUI structure to extract a set of key identifiers for each parameterized GUI widget. These identifiers are used to compose appropriate online search phrases and collect relevant test data from the Internet. The results of an empirical study on five GUI-based applications show that the proposed approach is applicable and results in execution of some hard-to-cover branches in the subject programs. The proposed technique works from a black-box perspective and is entirely independent from GUI modeling and event sequence generation, thus it does not require source code access and offers the possibility of being integrated with existing GUI testing frameworks.