Proposed measurement for complexity of analysis classes in UML

摘要

Several design metrics have been proposed for evaluating the complexity of object-oriented software. However, most of them discussed only the concepts or properties of the metrics, and have not shown any detailed procedure for computing the specific values. Some metrics requiring more detail than those currently provided by class definitions may well not be computable until implementation is complete and hence, are not practical design metrics. In this paper, we propose a new measurement based on class definitions that would be available during analysis. The metrics is aimed at evaluating the efforts required for understanding classes. It has been analytically evaluated using widely known properties for software metrics. As a result, the proposed metrics is found to satisfy the essential properties, that metrics must have. Finally, experimental evaluation of the metrics is also presented in the paper.