• 中国计算机学会会刊
  • 中国科技核心期刊
  • 中文核心期刊

计算机工程与科学

• 人工智能与数据挖掘 • 上一篇    下一篇

基于投射的藏语语义依存分析研究

夏吾吉1,2,华却才让1   

  1. (1.青海师范大学藏文信息处理教育部重点实验室,青海 西宁 810008;
    2.青海师范大学民族师范学院,青海 西宁 810008)
     
  • 收稿日期:2018-08-13 修回日期:2019-02-26 出版日期:2019-10-25 发布日期:2019-10-25
  • 基金资助:

    青海省科技计划项目(2017-GX-146);青海师范大学中青年科研基金(17ZR11)

Semantic dependence analysis of Tibetan based on projection

XIA Wu-ji1,2,HUAQUE Cai-rang1   

  1. (1.Tibetan Information Processing Key Laboratory of Ministry of Education,Qinghai Normal University,Xining 810008;
    2.Normal College for Nationalities,Qinghai Normal University,Xining 810008,China)
     
     
  • Received:2018-08-13 Revised:2019-02-26 Online:2019-10-25 Published:2019-10-25

摘要:

藏语是语序非常灵活的一种语言,藏语词法分析和句法分析等浅层研究不能很好地满足藏语自然语言理解的需求。从简单句型的藏语句子出发,研究了基于投射的藏语语义依存分析,构建了藏语语义依存树库,设计了语义依存弧类型分析特征模板。最后通过最大熵分类模型,对人工分析过的语义依存弧的句子进行依存弧的类型分析并进行标注,为今后的语义依存分析提供新的思考视角和更好的理论支撑。

 

关键词: 藏语语义, 投射, 语义依存树库, 最大熵模型

Abstract:

Tibetan is a language with very flexible word order. Current research such as Tibetan lexical analysis and syntactic analysis cannot meet the needs of Tibetan natural language understanding. Based on the simple sentence pattern of Tibetan sentences, we study the semantic dependence of Tibetan based on projection, construct a Tibetan semantic dependency tree library, and design a semantic dependency arc type analysis feature template. Finally, the semantically dependent arc sentences of the artificial analysis are analyzed and labeled through the maximum entropy classification model, which provides a new perspective and sound theoretical support for future semantic dependence analysis.
 

Key words: Tibetan semantics, projection, semantic dependent tree library, maximum entropy model