语义省略“的”字结构自动识别研究

Research on Automatic Recognition of“De”Construction with Semantic Ellipsis

语义省略是语言使用中存在的一类普遍现象,其省略的信息给机器自动理解造成困难。其中具有语义省略“的”字结构,在省略概念添加的类型中所占比例最高。文章利用“的”字局部上下文的词性和句法信息,通过动词框架找出具有语义省略的“的”字结构。实验表明,该方法能够在CTB8.0(Chinese Treebank)语料中有效识别出含有语义省略的“的”字结构,在测试集中F1值达到87%,取得了较好的实验效果,为机器对深层语义的理解奠定基础。

Semantic ellipsis is a common linguistic phenomenon in the use of language.The omission of information makes it difficult for the machine to understand semantics automatically.The“de”(的)construction with semantic ellipsis accounts for the highest proportion in the type of omitted concept addition.This paper uses the part-of-speech and syntactic information of the“de”local context and verb frame to find the de construction with semantic omission.Experiments show that this method can effectively identify the“de”construction with semantic ellipsis in CTB8.0(Chinese Treebank)corpus,and its F1 value reaches 87%,which achieves better experimental results and lays the foundation for machine understanding of deep semantics.