面向隐喻计算的语料库研究和建设

Corpus Designing and Constructing for Metaphor Computation

汉语隐喻计算化研究是中文自然语言处理领域的一个前沿课题,然而研究所需要的隐喻语料资源却极其有限,对隐喻计算研究的发展形成了一定的对影响。本文主要研究了面向隐喻计算的语料库的设计和构建方法,并设计构建了一个具有一万句规模62万字的汉语隐喻标注句库以及一个面向隐喻识别的汉语常用动词搭配库。这些资源的构建建立在对隐喻现象的深入分析基础之上,是隐喻计算模型研究的基础性资源,对汉语隐喻的认知计算有着积极意义。

The research of Chinese metaphor computation is one of the leading issues in the NLP field; however inadequate Chinese metaphor corpus limited development of it. This paper introduces the design and construction of corpus for metaphor computation; it constructs a Chinese metaphor corpus, which contains about 10,000 tagging sentences, 620,000 words, and a verb collocation corpus of common Chinese verbs. Construction of these corpuses is based on the in-depth analysis of the phenomenon of metaphor; they are the basic resources for the research of computational model of metaphor, and have a positive effect on Chinese cognitive computation of metaphor.