Abstract:

The results of software quality evaluation are closely related to users’ experience, but current software quality evaluation methods lack attention to user requirements because of the abstraction and complexity of software products and the fuzziness of customer requirements, thus ignoring the importance of user requirements on software quality evaluation. In consideration of the user requirements’ influence on the quality of software, a fuzzy neural network based on fuzzy numbers is proposed in order to imitate the nonlinear function between customer needs and software characteristics considering the influence of the user requirements on software quality, which can meet the complexity characteristics of software products, and make the results of software quality evaluation more objective and comprehensive. Results indicate that the method can more accurately imitate the nonlinear function between customer needs and software characteristics and is an effective measurement for studying software quality evaluation.

Key words: software quality;quality evaluation;fuzzy neural network;fuzzy numbers