从网页目录到轻量级本体——自然语言处理及形式化

From Web directories to lightweight ontology:natural language processing and formalization

为了解决以自然语言表示节点标签的分类树很难通过自动软件agents来进行自动推理的问题,通过词性标志、词义辨析、连接词辨析和受约束的自然语言定义及转换等步骤,将分类树中每一个节点对应的自然语言标签转换成了机器能够识别的逻辑表达式,从而使整个分类树转换成了一个轻量级本体,它适合应用在数据整合的语义匹配、文档分类和语义搜索等方面的自动推理,从而促进了本体知识的自动化推理,为以后文本自动检索奠定基础。

In order to solve the problem that classifications were very hard to be reasoned about by automated software agents and represent annotations of little use for semantic Web applications since their labels or nodes were written in natural language,this paper introduced an approach to transform a hierarchical directory into lightweight ontology by a series of steps,including part-of-speech tagging,word sense disambiguation,coordination disambiguation,and new controlled natural language definition and conversion,which then helped formalize the natural language labels into simple description logic formulae and provided the significant basis for further ontology reasoning and document retrieval.