• 中国计算机学会会刊
  • 中国科技核心期刊
  • 中文核心期刊

J4 ›› 2015, Vol. 37 ›› Issue (07): 1393-1398.

• 论文 • 上一篇    下一篇

藏文问答系统中问句的分类方法研究

柔特   

  1. (青海师范大学藏文信息处理省部共建教育部重点实验室,青海 西宁 810008)
  • 收稿日期:2014-01-13 修回日期:2014-09-01 出版日期:2015-07-25 发布日期:2015-07-25
  • 基金资助:

    国家自然科学基金资助项目(61163018,61262051,613630055);国家社科基金资助项目(14BYY132,13BYY141)

Research on question classification of
Tibetan question-answering system  

ROU Te   

  1. (Key Laboratory of Ministry of Education for Tibetan Information Processing,
    School of Computer Science,Qinghai Normal University,Xining 810008,China)
  • Received:2014-01-13 Revised:2014-09-01 Online:2015-07-25 Published:2015-07-25

摘要:

问句分类在问答系统领域有着重要的研究意义和应用价值。为了提高自动问句分类识别效率,本文提出了基于藏文疑问虚词的问句分类方法。该方法首先提取问句中的藏文虚词;然后根据所得的虚词进行问句分类;其次,进一步将问句细分为问题大类和小类;最后,给问题贴上分类标签。这种方法,既缩小了检索空间又提高了答案的准确性。最后对各种藏文问句分类的准确性进行了测试。结果表明基于藏文虚词问句的分类方法具有可行性和实用性。

关键词: 藏文, 问答系统, 问句分析, 问句分类, 疑问虚词, 特指问句

Abstract:

Question classification possesses significant research and application vaules in the questionanswering system. In order to improve the identification efficiency of automatic question classification, we present a new question classification method  based on the function words in Tibetan. We first extract Tibetan function words from the questions, and then classify the questions accordingly. Secondly, we further break the questions down into categories and subcategories, and finally affix the questions with different labels. The proposed method reduces the searching scope and improves the accuracy of the answers. We test the accuracy of Tibetan question classification function, and the results further validate that the question classification method based on Tibetan function words is feasible and practical.

Key words: Tibetan;question-answering system;question analysis;question classification;question function words;specific interrogation