首页> 外国专利> DEPENDENCY CONSTRAINT AND KNOWLEDGE-BASED ADJECTIVE MEANING DISAMBIGUATION METHOD AND APPARATUS

DEPENDENCY CONSTRAINT AND KNOWLEDGE-BASED ADJECTIVE MEANING DISAMBIGUATION METHOD AND APPARATUS

机译:依赖约束和基于知识的辅助测度消歧方法和装置

摘要

A dependency constraint and knowledge-based adjective meaning disambiguation method and apparatus, the method comprising: performing a dependency syntactic analysis on large-scale corpora, collecting the obtained dependency tuples and counting the absolute frequencies thereof, and creating a dependency knowledge library; performing a dependency syntactic analysis on sentences in which an ambiguous adjective is present, and extracting four dependency tuples meeting a set condition to be a dependency constraint set of the ambiguous adjective; according to a semantic dictionary, sequentially extracting, for respective meanings of the ambiguous adjective, a synonym set, a near-synonym set and an antonym set to be meaning representative word sets for corresponding meanings; according to the dependency knowledge library and the meaning representative word sets, sequentially calculating the posterior probability of each meaning of the ambiguous adjective in the dependency constraint set; and selecting an accurate meaning of the ambiguous adjective according to the posterior probability. By utilizing the method and the apparatus, the effect of a dependency syntactic analysis can be brought into full play and the meaning of an ambiguous adjective can be determined more accurately and effectively.
机译:一种依赖关系约束和基于知识的形容词义消歧方法和装置,该方法包括:对大型语料库进行依赖关系句法分析,收集获得的依赖关系元组并计算其绝对频率,并建立依赖关系知识库。对存在歧义形容词的句子进行依存句法分析,提取满足设定条件的四个依存元组作为歧义形容词的依存约束集;根据语义词典,针对歧义形容词的各个含义,依次提取同义词集,近义词集和反义词集,作为对应含义的意义代表词集;根据依赖关系知识库和含义代表词集,依次计算依赖关系约束集中歧义形容词每种含义的后验概率;并根据后验概率选择歧义形容词的准确含义。通过使用该方法和装置,可以充分发挥依存句法分析的作用,可以更准确,更有效地确定歧义形容词的含义。

著录项

  • 公开/公告号WO2018001101A1

    专利类型

  • 公开/公告日2018-01-04

    原文格式PDF

  • 申请/专利权人 QILU UNIVERSITY OF TECHNOLOGY;

    申请/专利号WO2017CN88364

  • 发明设计人 LU WENPENG;

    申请日2017-06-15

  • 分类号G06F17/27;

  • 国家 WO

  • 入库时间 2022-08-21 12:46:39

相似文献

  • 专利
  • 外文文献
  • 中文文献
获取专利

客服邮箱:kefu@zhangqiaokeyan.com

京公网安备:11010802029741号 ICP备案号:京ICP备15016152号-6 六维联合信息科技 (北京) 有限公司©版权所有
  • 客服微信

  • 服务号