【摘要】 提出一种基于知网和术语相关度的关系抽取方法。首先通过句法分析提取术语的上下文特征,结合自然语言特征和互信息的方法计算术语之间的相关度,然后使用术语的义原和动态角色作为关键词,在知网语义关系框架中定位关系,并为关系指定明确的语义标签。实验结果表明该方法具有较好的实用效果。更多还原

【Abstract】 The paper proposes a relationship extraction method based on HowNet and term relevancy degree.Firstly syntax parsing tools are used to extract context feature of terms,and natural language feature and statistical mutual information measure are integrated to compute relevancy degree of terms,then dynamic role and sememe are used as key to seek the relationship in HowNet semantic relationship framework,and explicit semantic lable is designated to the relationship.Experimental results show that the approach is effective.更多还原