一种基于h指数的软件网络中重要类的度量方法

Approach Based h-index to Measuring the Important Classes in Software Network

在软件开发过程中,为加快新开发者对系统的了解,往往优先推荐他们关注一些更重要的类,从而快速掌握系统最基本的概念.目前,关于网络节点重要性度量的方法很多,h指数作为一种成功用于定量评估研究人员学术成就的指标却很少应用于软件网络中重要类的识别.以Tomcat、Ant和Jung项目为研究对象,使用h指数来度量软件系统中类的重要性,并与已有的中心性指标进行对比,验证该指标的可行性以及评估的效果.实验结果表明,h指数能够很好的识别出软件系统中更重要类,尤其是考虑节点的连边权重时,且相比已有的中心性指标效果更好.

To improve the newcomers＇ understanding of a software system during its development, the important classes are in general given priority to be recommended to the newcomers who need to master some basic concepts as soon as possible. At present, there are many methods for measuring the importance of node in a network. As a metric successfully applied to evaluate the productivity of a scholar,h-index is merely used to identify the important classes in software network. In this paper, we attempted to use h-index to measure the important classes on three open-source software systems （i. e. , Tomcat, Ant, and Jung ）, and validated the feasibility and experimental effect of h-index by comparing the result with that obtained from using several existing centrality metrics. The results show that h-index proposed in this paper can identify those important classes, in particular the h-index considering the weight of each edge performs better in this problem.