A Novel Mask-Coding Representation for Set Cover Problems with Applications in Test Suite Minimisation

Abstract

Multi-Objective Set Cover problem forms the basis of many optimisation problems in software testing because the concept of code coverage is based on the set theory. This paper presents Mask-Coding, a novel representation of solutions for set cover optimisation problems that explores the problem space rather than the solution space. The new representation is empirically evaluated with set cover problems formulated from real code coverage data. The results show that Mask-Coding representation can improve both the convergence and diversity of the Pareto-efficient solution set of the multi-objective set cover optimisation.