中文专利领域本体概念间非分类关系抽取  

Extract non-taxonomic relations between ontological concepts from Chinese patent documents

作　　者：何宇[1] 吕学强[1] 刘秀磊[2] 徐丽萍[3] 

机构地区：[1]北京信息科技大学网络文化与数字传播北京市重点实验室,北京100101 [2]北京信息科技大学计算机学院,北京100101 [3]北京城市系统工程研究中心,北京100089

出　　处：《计算机工程与设计》2017年第1期97-102,共6页Computer Engineering and Design

基　　金：国家自然科学基金项目(61271304);北京市教委科技发展计划重点项目暨北京市自然科学基金B类重点基金项目(KZ201311232037);北京市属高等学校创新团队建设与教师职业发展计划基金项目(IDHT20130519);北京市科研院创新工程基金项目(PXM2015_178215_000008)

摘　　要：将中文专利本体中实体间非分类关系抽取问题限定为满足SAO结构的实体间关系抽取问题。针对SAO结构关系抽取任务中关系实例结构正确但是语义错误的问题,提出一种句法分析特征和关系词词典特征结合传统特征的关系抽取方法。将新特征和上下文特征、距离特征等传统特征相结合,使用支持向量机进行关系抽取实验。实验结果表明,该方法优于仅使用传统特征的关系抽取方法。The non-taxonomic relations between concepts in Chinese patent ontology were limited to the problem of SAO-based relation extraction from Chinese patent documents. To solve the problem that relation structure is right but the semantic is wrong, syntactic information and dictionary of relationship were proposed to use. Support vector machine was used and traditional features such as lexical information, distance of concepts were combined with new features to carry on the experiments. Results show that this method has better performance than traditional one.

关 键 词：SAO结构 非分类关系抽取 句法分析 支持向量机 本体学习 

分 类 号：TP391.1[自动化与计算机技术—计算机应用技术]