A New Test Suite Reduction Approach Based on Hypergraph Minimal Transversal Mining

摘要

Test Suite Reduction (TSR) approaches aim at selecting ordy those test cases of a test suite to reduce the execution time or decrease the cost of regression testing. They extract the tests that cover test requirements without redundancy, or exercise changed parts of the System Under Test (SUT) or parts affected by changes, respectively. We introduce DTSR (Deterministic Test Suite Reduction), that relies on the hypergraph structural information to select the candidate test cases. Requirement data, which are associated with the test cases, optimize the selection by retaining a deterministic set. To do so, DTSR considers a test suite as a hypergraph, where its nodes are equivalent to tests, and its hyperedges are similar to requirements. The algorithm extracts a subset of the minimal transversals of a hypergraph by selecting the minimum number of test cases satisfying the requirements. We compare our new algorithm versus search based ones, and we show that we outperform the pioneering approaches of the literature. The reduction rate varies from 50% up to 65% of the initial set size.