决策树分类方法在软构件检索中的辅助决策支持研究

Research on decision-tree classification in component retrieval auxiliary decision support

构件库的检索效率不仅包括对构件本身的检索效率,还包括构件查询者对构件的理解效率。而一般的构件库系统只是从描述性信息方面提供对构件的理解,却很少从实际复用方面提供对构件的理解。数据挖掘技术为解决上述问题提供了一条可行的途径,文中对如何利用决策树分类方法进行可复用构件的复用历史信息和用户反馈信息隐含知识的挖掘作一探讨。通过对数据挖掘技术的应用,使构件库的相关人员能够从其他复用者实际复用构件的角度来理解构件,从而为构件生产者改进构件、管理者管理构件、复用者理解和选取构件提供一定程度的辅助决策支持。最后,通过实验验证了这种方法的可行性与有效性。

The retrieval efficiency of the component library includes not only the retrieval efficiency of the component itself, but also the efficiency of the understanding of component. General component retrieval system provides descriptive information on understanding the component simply, but seldom provides information from the aspect of reusing actually. Data Mining technologies provide a feasible approach for effective decision support. In this paper, the application of decision-tree-based classification method was discussed. It could excavate the cryptic knowledge in component user feedback information, and enable the relevant personnel in the component library to understand the component from the aspect of using the component actually. DM solutions improved possibility and quality of reusing. In the end, the feasibility and validity of this method were verified through experiments.