Applying Elementary Landscape Analysis to Search-Based Software Engineering

Abstract

Recent research in search-based software engineering (SBSE) has demonstrated that a number of software engineering problems can be reformulated as a search problem, hence search algorithms can be applied to tackle it. However, most of the existing work has been of empirical nature and the techniques are predominately experimental. Therefore in-depth studies into characteritics of SE problems and appropriate algorithms to solve them are necessary. In this paper, we propose a novel method to gain insight knowledge on a variant of the next release problem (NRP) using elementary landscape analysis, which could be used to guide the design of more efficient algorithms. Preliminary experimental results are obtained to indicate the effectiveness of the proposed method.