Induction of Concepts in Web Ontologies through Terminological Decision Trees
Induction of Concepts in Web Ontologies through Terminological Decision Trees
复制标题
通过术语决策树归纳网络本体中的概念
DOI：
--
复制
发表时间：
2010
期刊：
影响因子：
--
通讯作者：
F. Esposito
中科院分区：
文献类型：
--
作者：
N. Fanizzi;Claudia d’Amato;F. Esposito
提出了一种新的逻辑决策树归纳框架。与原始设置不同，树节点上的测试用描述逻辑概念表示。这有许多优点：可表达的术语语言被赋予完全否定，从而允许在每个测试节点上更自然地划分个体；这些逻辑支持用于在语义Web中表示知识库的标准本体语言。提出了一种自顶向下的术语决策树生成方法，该方法是对著名的树归纳方法的改进。这为在描述逻辑中学习提供了另一种方法，因为概念描述可以与术语树相关联。实现了这些方法的新版本的系统TermiTIS在来自流行知识库的本体上进行了实验评估。
A new framework for the induction of logical decision trees is presented. Differently from the original setting, tests at the tree nodes are expressed with Description Logic concepts. This has a number of advantages: expressive terminological languages are endowed with full negation, thus allowing for a more natural division of the individuals at each test node; these logics support the standard ontology languages for representing knowledge bases in the Semantic Web. A top-down method for inducing terminological decision trees is proposed as an adaptation of well-known tree-induction methods. This offers an alternative way for learning in Description logics as concept descriptions can be associated to the terminological trees. A new version of the System TermiTIS, implementing the methods, is experimentally evaluated on ontologies from popular repositories.