Evolutionary algorithms for the multi-objective test data generation problem

摘要: Automatic test data generation is a very popular domain in the field of search-based software engineering. Traditionally, main goal has been to maximize coverage. However, other objectives can be defined, such as oracle cost, which cost executing entire suite and checking system behavior. Indeed, large systems, spent an issue, then it makes sense by considering two conflicting objectives: maximizing coverage minimizing cost. This what we did this paper. We mainly compared approaches deal with multi-objective problem: direct approach combination mono-objective algorithm together case selection optimization. Concretely, work, used four state-of-the-art algorithms evolutionary followed based on Pareto efficiency. The experimental analysis compares these techniques different benchmarks. first one composed 800 Java programs created through program generator. second benchmark 13 real extracted from literature. In approach, results indicate that properly optimized; however, full branch poses great challenge. Regarding algorithms, although they need phase for reducing are effective Copyright © 2011 John Wiley & Sons, Ltd.