韩国语名词短语结构特征分析及自动提取

Structure Characteristic Analysisand Automatic Extraction for Korean Noun Phrase

名词短语作为语言中一种普遍的语法现象,在自然语言处理领域日益受到了研究人员的关注。目前,对其研究范围主要集中在边界识别、语法分析、语义分析及其分类等方面。该文通过研究分析韩国语书面语名词短语的左右边界规则,从大规模标注语料库中自动提取出名词短语。实验结果表明:语料中的高频名词短语相对集中于8个类型之中。根据提取结果分别建立不同类型的名词短语库,为进一步建立双语平行短语语料库打下基础,以便于以后的机器翻译、信息检索等自然语言信息处理工作。

These years, noun phrase,as a common grammatical phenomenon,has attracted eyes of many scholars in the field of language processing.At present,most researches on noun phrase lie in boundary identification grammatical analysis,semantic analysis,categorization and some other aspects.This thesis abstractsnoun phrases from a large-scale tagged corpus through studying and analyzing rules of left and right boundaries of noun phrases in written Korean.From the experimental result,we can see that high-frequency noun phrases mainly lie in 8 categories.Different kinds of corpus for noun phrases can be built according to the result of the abstract,which lays the foundation of building paralell corpus.It will also be convenient for machine translation,information retrieval and other work in language information processing in the future.