期刊论文

Inner-Attention Based Multi-Way Convolutional Neural Network for Relation Recognition in Chinese Compound Sentence

中文

中文信息学报

1003-0077

2020

34

6

9-17,26

10.3969/j.issn.1003-0077.2020.06.003

国家社会科学基金(18BYY174) 教育部人文社会科学研究规划基金(14YJA740020)

本校为第一机构

复句的关系识别是对分句间语义关系的甄别,是复句语义分析的关键,旨在从文本中识别句间的关系类型.非充盈态汉语复句存在隐式关系的特点给语义关系识别造成了困难.为了深度挖掘复句中隐含的语义信息,正确地实现关系分类,该文提出了一种基于句内注意力机制的多路CNN网络结构Inatt-MCNN.其中句内注意力机制模型是基于Bi-LSTM的,使其能够学习到句子的双向语义特征以及分句间的关联特征.同时,为了充分利用文本特征,联合使用卷积神经网络(CNN)对复句表示再次建模获得句子局部特征.与其他基于汉语复句语料库(CCCS)和清华汉语树库(TCT)的实验结果相比,该文方法的宏平均F 1值为85.61％,提升约6.08％,平均...

Compound sentence relation recognition is to identify for the semantic relation of clauses, which is the key task in semantic analysis of compound sentences. This task is difficult due to the an implicit relation in non-saturate compound sentences. To deal with the implicit semantic information, a multi-channel CNN based on the inner-attention mechanism is proposed in this paper. The inner-attention mechanism is based on Bi-LSTM, which enables it to learn bidirectional semantic features and associated features between clauses. At the same time, CNN is used to model the sentence representation ...