计算机工程与应用 ›› 2015, Vol. 51 ›› Issue (17): 107-112.

• 数据库、数据挖掘、机器学习 • 上一篇    下一篇

基于事件语义距离的V1-V2述结式判别研究

马  腾,詹卫东   

  1. 北京大学 中文系,中国语言学研究中心,计算语言学教育部重点实验室(北京大学),北京 100871
  • 出版日期:2015-09-01 发布日期:2015-09-14

Identification of verb-resultative construction based on event distance

MA Teng, ZHAN Weidong   

  1. Department of Chinese Language & Literature, Center for Chinese Linguistics PKU, Key Lab of Computational Linguistics, MoE, Peking University, Beijing 100871, China
  • Online:2015-09-01 Published:2015-09-14

摘要: “现代汉语述补结构用法词典”是人工建立、用于描述述补结构相关信息的语言知识资源。经过人工对词条的收集、释义等编写工作,词典已形成一定规模。在此基础上,尝试借助计算机技术,依据事件语义学的理论,利用现有语言知识资源以及大规模语料数据,寻找述结式复合事件语义距离计算的方法,对述结式进行定量描写,以帮助扩大词典规模,同时有助于深化对特有语言现象——述补结构的认识。实验结果表明该方法具有较高的准确率和识别率。

关键词: 述结式, 复合事件, 语义距离

Abstract: The Dictionary  of Verb-Resultative Construction in Contemporary Chinese is created by hand, to describe the knowledge about verb-resultative construction. Based on this dictionary and the big scale of corpus data, it proposes an algorithm to calculate the event distance in a V1-V2 verb-resultative construction. With this model, it can identify the verb-resultative construction from different V1-V2 combination. Using this method, it can extract the V1-V2 verb-resulative constructions from corpus automatically with high accuracy.

Key words: verb-resultative construction, compound event, event distance