Web超链挖掘:中国境内Web图结构研究  

Website Linkage Mining and Web Graph Structure of China

作　　者：丁国栋[1] 王斌[1] 白硕[1] 

机构地区：[1]中国科学院计算技术研究所,北京100080

出　　处：《计算机工程》2005年第14期24-26,共3页Computer Engineering

摘　　要：以网站作为Web图的顶点,以网站之间链接为有向边,研究了中国境内Web图的拓扑特点和宏观结构。试验表明:网站的入度和出度分布同样服从幂级数定律(PowerLaw);境内Web图的连通性明显高于全球的Web图,其最大的强连通分量中的网站数超过50%;在境内Web中,如果两个网站之间存在一条有向路径,则从一个网站漫游到另外一个网站,平均只需点击7.1次,最多只需点击29次。Previous studies of the Web graph structure focuses on the graph at the level of individual Web pages, which suggest the independency hypothesis between pages. To better understand the macro-structure of the Web graph, this paper proposes a new method at the level of websites. The experiments show that: power laws also arise in the in-and out-degree distributions of websites; the Web of China has an obviously higher connectivity than the global Web; on the Web of China, the average shortest-path-length is 7.1 and the maximum is 29.

关 键 词：Web图 网站 链接关系 宽度优先搜索 强连通分量 

分 类 号：TP393[自动化与计算机技术—计算机应用技术]