Research And Application On The Metrics Of Object-Oriented Software Quality

Along with the application area of software becoming more and more widespread, the quality of software has become more serious to people. With the Object-Oriented technology becoming widespread, the importance of Object-Oriented metrics should not be disregarded. Furthermore, if we can predict the complexity of software at the beginning of software development, we could make people who are engaged in software work better.Firstly, this thesis summarily depicted the development of software metrics, introduced frameworks and models of software metrics, analyzed and summarized the structure-oriented metrics.In the following, the trait of Object-Oriented software was analyzed, and some popular Object-Oriented design metrics were investigated and analyzed, such as C&K metrics suits and MOOD metrics etc. The limitation of them, especially the C&K metrics was summarized; further investigation and complement of the metrics had been done, application methods of C&K metrics was analyzed too.At last, this thesis analyzed and compared current popular UML-based metric model, and pointed out the defects of them. Then from practice, an experiment was done according to GQM(Goal-Question-Metric) which was proposed by Basili. Through collecting and analyzing the data from experiment, applying multivariate linear equation and the theory of statistics a few alternative quality prediction models which were based UML class diagram were proposed, and concerning measurement metrics were defined fully. Through analyzing and comparing these models, two accurate models were derived. Through studying the two models, it was discovered that the software quality prediction model could be built on the basis of two measurement metrics of UML class diagram. Through applying the two models, the quality of the software could be predicted well.