Bridging the model-to-code abstraction gap with fuzzy logic in model-based regression test selection

Abstract Regression test selection (RTS) approaches reduce the cost of regression testing evolving software systems. Existing RTS based on UML models use behavioral diagrams or a combination structural and diagrams. However, in practice, are incomplete not used. In previous work, we proposed fuzzy logic approach called FLiRTS that uses sequence activity this introduce 2, which drops need for relies system only class diagrams, most widely used practice. 2 addresses unavailability by classifying cases using after analyzing information commonly provided We evaluated extracted from 3331 revisions 13 open-source systems, compared results with those code-based dynamic (Ekstazi) static (STARTS) approaches. The average suite reduction was 82.06%. safety violations respect to Ekstazi STARTS were 18.88% 16.53%, respectively. selected about 82% STARTS. precision 13.27% 9.01%, mutation score full suites 18.90%; standard deviation reduced each subject 1.78% 1.11% Ekstazi, 1.43% Our experiment demonstrated performance is close state-of-art tools but requires less performs time.