【作者】 孙俊； 曹海龙； 赵铁军；

【Author】 Jun Sun, Hailong Cao, Tiejun Zhao (School of Computer Science and Technology, Harbin Institute of Technology, Harbin 150001)

【机构】 哈尔滨工业大学计算机学院；

【摘要】 句法分析是自然语言处理的一个基本问题,也是目前急待解决的一个问题。目前大多数的句法分析是基于统计方法的,基于统计的句法分析需要大规模的训练语料,而标注一个大规模语料需要很大的人力。为了减少标注句法树库所需的人力, 本文对选择样本进行了研究。本文从句法结构上对句子进行聚类,根据聚类的结果精选出一个小的句子集,这个句子集的规则分布近似于整个句子集的规则分布。标注这个句子集就能在保证句法分析器性能的前提下减少标注所需的人力。实验结果证明, 通过选取一半的句子训练出的句法分析器,其性能就能近似于用所有句子训练的句法分析器的性能。更多还原

【Abstract】 Parsing is one of the fundamental problems in natural language processing, and the main approach is statistical parsing. Statistical parser relies on using many hand-parsed sentences as training examples. However, the task of labeling so many sentences is a labor-intensive task. We proposed to select these samples to reduce the amount of sentences in the training data by clustering based on syntactic structures, thereby reducing the workload of human to label sentences. Our result showed that the amount of training sentence could be reduced by 50% with an approximate performance of the statistical parser.更多还原