基于语义依存的中文本体非分类关系抽取方法

摘要

To deal with Chinese ontology non-taxonomic relation extraction, an approach for extraction based on semantic dependency is proposed. Firstly, semantic role labeling and dependency grammatical analysis is used to obtain semantic role and dependency relation of text sentence element. Verb frame with semantic dependency from text is extracted Then, non-taxonomic relation between concepts and name of the relation are discovered by computing semantic similarity. Experiments show that the proposed approach realize non-taxonomic relation extraction and semantic labeling for relation effectively.%为了解决中文本体非分类关系抽取问题,提出了基于语义依存分析的非分类关系抽取方法.利用语义角色标注和依存语法分析思想,分析得到了文本句子的语义依存结构,提取其中具有语义依存关系的动词框架,通过计算语义相似度,发现了动词框架中概念间的非分类关系和关系名称.实验结果表明该方法能够有效地实现非分类关系的抽取和关系的语义标注.