Automatic Extraction of Behavioral Features for Test Program Similarity Analysis

We present a methodology for performing automatic extraction of behavioral features from test programs, that is, for collecting pieces of information about the test programs execution. These features are then exploited to carry out analysis and reasoning about test program similarity. The similarity information can be used to drive the execution of test campaigns, in the attempt to either reduce the time-to-test, or to increase the testing capabilities of a given test suite. Our methodology is embedded in the Hyperion analysis framework, which can be configured to define a wide range of test program similarity criteria.