基于Markov逻辑网的超文本分类  

Hypertext classification based on Markov logic networks

作　　者：张玉芳[1] 孔润[1] 田源[1] 熊忠阳[1] 

机构地区：[1]重庆大学计算机学院,重庆400044

出　　处：《南京大学学报（自然科学版）》2011年第5期571-577,共7页Journal of Nanjing University（Natural Science）

基　　金：重庆市自然科学基金(CSTC2008BB2191)

摘　　要：在传统的监督学习任务中,实体被认为是独立同分布的.然而,现实世界中实体之间通过复杂的方式相互关联.例如在超文本分类中,具有链接关系的页面之间高度相关.标准的分类方法是忽略实体之间的联系,对每个实体单独分类.本文将Markov逻辑网应用到超文本分类中,旨在改善这一问题.实验结果显示了采用Markov逻辑网模型要比采用K最邻近节点算法的分类效果好;同时将实体之间存在的联系用于学习和推理对于分类也有一定的贡献.In traditional supervised learning tasks,the labeled entities are related to each other in complex ways and their labels are not independent.For example,in hypertext classification,the labels of linked pages are highly correlated.A standard approach is to classify each entity independently,ignoring the correlations between them.We use a statistically relational learning model,Markov logic networks,in hypertext classification in order to solve this problem.Our experiments prove that this model has better performance than k-nearest neighbor does in hypertext classification and the correlations between the entities benefit for the performance as well.

关 键 词：MARKOV逻辑网 超文本分类 MARKOV网 统计关系学习 

分 类 号：TP391[自动化与计算机技术—计算机应用技术]