首页|知识元逻辑关系抽取方法研究

知识元逻辑关系抽取方法研究

扫码查看
知识元逻辑关系抽取关注上下文的句法结构与语法特征,通过上下文功能语义依赖定义触发词规则实现上下文范围内的关系抽取.为摆脱上下文的结构限制,本文提出一种知识元逻辑关系抽取的新方法.首先,面向领域文献资料,通过知识元抽取构建知识元集合;其次,基于领域知识特征构建知识元属性描述框架,对知识元进行全方位、细粒度的统一描述,基于知识元属性抽取构建知识元属性集合;最后,结合实际需求定义逻辑关系类型,采用逻辑关联实例分析与属性对比关联分析,以属性值的共现、包含、相关等基本关系为线索,通过特征分析与归纳总结构建逻辑关系规则库,通过规则匹配实现知识元逻辑关系抽取.以南海维权证据知识元为例,选取部分不同文本来源的证据,定义了证据间逻辑关系类型及其抽取规则,在证据知识元及其属性抽取的基础上,基于规则匹配实现南海维权证据间并列、承接、反驳与补强4种逻辑关系的抽取,构建了南海维权证据逻辑关联图谱.实证研究验证了本文方法的可行性,能够为知识元逻辑关系抽取的相关研究提供可参考的新思路.
Knowledge Element Logical Relation Extraction Method
The extraction of knowledge element logical relation focuses on the syntactic structure and syntactic features of the context and realizes context-scoped relation extraction by defining trigger word rules through contextual functional se-mantic dependencies.To eliminate the structural limitation of context,this study proposes a new method for the extraction of knowledge element logical relation.First,based on the domain literature,the knowledge element collection is construct-ed through knowledge element extraction.Second,based on the domain knowledge characteristics,a knowledge element attribute description framework is constructed to provide an all-round and fine-grained unified description of knowledge el-ements,and the knowledge element attribute collection is constructed based on knowledge element attribute extraction.Fi-nally,the logical relation types are defined in accordance with the actual needs.Through logical relation instance analysis and attribute comparative relation analysis,the logical relation rule base is constructed via characterization and summariza-tion.This is done by taking the basic relations,such as co-occurrence,inclusion,and correlation of attribute values,as clues,and the logical relation extraction is realized through rule matching.Considering the South China Sea rights protec-tion evidence knowledge element,some evidence from different text sources are selected,and the types of logical relations between evidence and their extraction rules are defined.Based on the extraction of knowledge elements and their attri-butes,the logical relation extraction between the South China Sea rights protection evidence of juxtaposition,succession,rebuttal,and reinforcement is realized based on rule matching,and the logical relation mapping of the South China Sea rights protection evidence is constructed.The feasibility of the proposed method is empirically verified,which can provide new ideas for research on the logical relation extraction of knowledge elements.

knowledge elementlogical relationrelation extractionthe South China Sea rights protectionevidence knowledge element

程为、郑德俊、朱梦蝶、丛天时、王燕红

展开 >

南京农业大学信息管理学院,南京 210095

南京大学信息管理学院,南京 210023

知识元 逻辑关系 关系抽取 南海维权 证据知识元

国家社会科学基金重大项目

19ZDA347

2024

情报学报
中国科学技术情报学会 中国科学技术信息研究所

情报学报

CSTPCDCSSCICHSSCD北大核心
影响因子:1.296
ISSN:1000-0135
年,卷(期):2024.43(7)
  • 28