摘要
基于链接的特点,提出了获取链接信息的模型,将得到的链接信息结合对象本身的属性来共同训练分类规则。针对网页链接的特殊性,对链接有向图重新建模。实验证明链接信息的加入可以有效地改善分类的结果,链接有向图的重新建模同样提高了分类的准确性。
Based the properties of link, this paper proposed a model to gain the link information. Then combined link information with the object attributes in order to train the rule of classification. Because of the particularity of Web pages, remodeled the graph which formed according to the link structure. The experiments show that using of the link information can improve the result of classification, and after the remodel of the graph, the result of classification is improved too.
出处
《计算机应用研究》
CSCD
北大核心
2008年第11期3271-3274,共4页
Application Research of Computers
基金
国家自然科学基金资助项目(60673136)
关键词
链接挖掘
基于链接的分类
逻辑回归模型
link mining
link-based classification
logistic regression model
