This research presents a new test suite consolidation technique, called CONTEST, for automated GUI testing. A new probabilistic model of the GUI is developed to allow direct application of CONTEST. Multiple existing test suites are used to populate the model and compute probabilities based on the observed event sequences. These probabilities are used to generate a new test suite that consolidates the original ones.

A new test suite similarity metric, called CONTeSSi(n), is introduced which compares multiple event sequence-based test suites using relative event positions. Results of empirical studies showed that CONTEST yields a test suite that achieves better fault detection and code coverage than the original suites, and that the CONTeSSi(n) metric is a better indicator of the similarity between sequence-based test suites than existing metrics.