Ontology自动创建中实例学习的研究  

Instance Learning in Automatic Construction of Domain Ontology

作　　者：刘贺欢[1] 刘椿年[1] 

机构地区：[1]北京工业大学计算机学院多媒体与智能软件技术北京市重点实验室,北京100022

出　　处：《计算机应用研究》2006年第3期38-40,共3页Application Research of Computers

基　　金：国家自然科学基金资助项目(60173014);北京市自然科学基金资助项目(4022003)

摘　　要：Ontology近年来受到信息科学领域的广泛关注,其重要性已在许多方面表现出来并得到广泛认同。自动创建领域Ontology可以克服手工方法的不足,成为当前的研究热点之一;实例(Instance)是Ontology的重要组成元素,从领域文档集中学习实例是自动创建领域Ontology的关键之一。研制的一个领域Ontology的自动生成系统(OntoAGS)能够通过领域文档集自动地创建该领域的Ontology,OntoAGS系统的实例学习是基于模式匹配的算法。实验表明,与当前较流行的Ontology半自动生成系统text-to-onto相比,OntoAGS通常能学习出更多实例,且准确率较高。The importance of Ontology is being recognized in various research and application areas. Constructing domain Ontology takes too much time and manpower. Hence, it is of great significance to do some research on constructing domain Ontology automatically. Instance is an important element of Ontology. Learning instance from domain corpus is a key issue in automatic construction of domain Ontology. A domain Ontology construction system OntoAGS has been developed. Given some domain corpus, OntoAGS can construct domain Ontology automatically. The instance learning method of this system is based on pattern matching algorithm. Experiments show that OntoAGS can generally learn more instances and more accurately than text-to-onto system which is a popular Ontology semi-automatic construction system.

关 键 词：自动创建领域Ontology 模式匹配 实例学习 

分 类 号：TP39[自动化与计算机技术—计算机应用技术]