基于HowNet的中文本体学习方法研究

摘要

目前针对国内在中文环境下本体学习的研究才刚刚起步的现状,对本体学习和HowNet进行了简单介绍,提出了基于HowNet的中文本体学习的主要思路.当前,本体学习的研究重点在于概念及概念间关系抽取.采用文本语料作为输入,首先对文本进行预处理,然后基于HowNet生成了一个领域语义词典,在本体学习中加入领域核心概念本体,在概念关系抽取阶段,采用基于HowNet的语义相似度计算方法.实验证明,提出的本体学习方法能够有效改进概念和概念间关系抽取的准确度.%At present based on ontology learning just started in the environment of Chinese, give a brief introduction about HowNet and ontology learning ,and propose the main ideas based on the Chinese HowNet. Ontology learning is focused on the concept extraction and relation extraction recently. The text corpus as input, the first of the pre-text,then it generates a field dictionary based on HowNet semantic , and adds the core concepts ontology into the field of ontology learning , in the concept relation extraction stage, the similarity calculation is based on HowNet methods. Experimental results show that the proposed method can effectively improve the accuracy of ontology concept extraction and relation extraction.