Abstract: This paper improves predictive object points method and proposes a UML based metric to solve the problems of software estimation accuracy. Case model, sequence diagram and class diagram can be used to get software metric. The result proves this method’s precision increases by 5% compared with predictive object points, and subsystems have same deviation direction.

Key words: software metric, software engineering, UML, predictive object points

摘要： 针对现有软件度量估算困难、精确度不高的问题，在预测性对象点基础上改进并提出基于UML的软件度量方法，可以从用例、序列图、类图度量软件。结果证明，与预测性对象点相比，该方法的精度提高了5%左右，其子系统偏差方向一致。

关键词: 软件度量, 软件工程, 统一建模语言, 预测性对象点

CLC Number: 

• TP311