多重分割框架下的两类新本体学习算法  

Two New Classes of Ontology Learning Algorithms in Multi-Dividing Setting

作　　者：朱林立 ZHU Lin-li(School of Computer Engineering,Jiangsu University of Technology,Changzhou 213001,China)

机构地区：[1]江苏理工学院计算机工程学院,江苏常州213001

出　　处：《云南师范大学学报（自然科学版）》2020年第3期34-39,共6页Journal of Yunnan Normal University:Natural Sciences Edition

基　　金：国家自然科学基金资助项目(51574232)。

摘　　要：多重分割本体学习算法的本质是将整个一维实数轴分割成k段,并依次分配给本体图上的k个分割类的顶点.在具体学习过程中,对不符合预定规则的本体样本对进行惩罚,进而得到最优本体函数.给出两类多重分割框架下的本体学习算法,并且利用统计学习理论中覆盖数逼近的方法对第一类本体学习算法进行理论分析.The essence of the multi-dividing ontology learning algorithm is to divide the entire one-dimensional real number axis into k segments and assign them to the vertices of k classes on the ontology graph.In the specific learning process,ontology sample pairs that do not meet the predetermined rules are punished,and then obtain the optimal ontology function.This paper presents two ontology learning algorithms in multi-dividing setting,and uses the trick of covering number approximation in statistical learning theory for theoretical analysis of the first type of ontology learning algorithm.

关 键 词：本体 机器学习 多重分割 覆盖数 

分 类 号：TP391[自动化与计算机技术—计算机应用技术]