Prioritization of Abstract Test Cases by Weighting Covered Value Combinations

Abstract Test Cases (ATCs) have been widely used in practical scenarios such as combinatorial testing. When the set of ATCs is constructed, Test Case Prioritization (TCP) is applied to improve the testing quality. Among the many TCP approaches for ATCs, one intuitive and well-studied approach is Interaction Coverage Based Prioritization (ICBP), which greedily identifies one element as the next ATC such that it covers the largest number of uncovered $\tau$-wise parameter-value combinations. When a $\tau$-wise value combination, $\mathrm{v}\mathrm{c}$, has been covered by the already-selected ATCs, it is generally believed that it may have no further role to play in the ICBP process. However, each already-covered $\mathrm{v}\mathrm{c}$ may provide some potential contributions for the interaction coverage of the strength higher than $\tau$. Motivated by this, we propose a new TCP approach to prioritize ATCs, Weighted Interaction Coverage Based Prioritization (WICBP). WICBP attempts to assign a weight to each $\tau$-wise value combination that has already been covered. The experimental results show that when using the same prioritization strength, WICBP is more effective than ICBP but maintains comparable testing efficiency. In addition, WICBP with lower prioritization strengths achieves surprisingly competitive (even better) performance compared with ICBP with higher prioritization strengths.