一种利用度属性获取本体概念层次的方法  

Concept Hierarchy Induction Approach Based on Degree

作　　者：温春[1] 石昭祥[1] 杨国正[1] 

机构地区：[1]电子工程学院网络工程系,安徽合肥230037

出　　处：《小型微型计算机系统》2010年第2期322-326,共5页Journal of Chinese Computer Systems

基　　金：电子工程学院博士创新基金(2008006)资助

摘　　要：通过分析已有的中文本体概念层次获取方法的特点和不足,提出一种利用度属性获取概念层次的方法.首先将概念关系以图的形式表示出来,然后利用反向剪枝算法得出概念所在层次,最后补充并修剪层次关系生成完整的概念层次.从各种参数的对比结果来看,基于VSM(Vector Space Model)且不进行LSI(Latent Semantic Indexing)降维时效果最好,同时也优于目前已有的几种方法.By analyzing the characteristics and disadvantages of several Chinese ontology concept hierarchy induction methods,this paper proposes an induction method based on degree. Firstly the concept hierarchy relation is presented as a graph. Then the hierarchies of concepts are exploited with a reverse pruning algorithm. At last,after adding and pruning the concept relation the concept hierarchy is generated. Compared with variable value of parameter,the best performance,which is better than that of other methods,is reached when based on VSM but not using LSI.

关 键 词：概念层次 本体学习 度 反向剪枝 

分 类 号：TP182[自动化与计算机技术—控制理论与控制工程]