计算机与数字工程2025，Vol.53Issue(4)：995-1001,7.DOI:10.3969/j.issn.1672-9722.2025.04.014
基于词性特征和混合神经网络的中文作文优美句识别
Elegant Sentences Recognition Based on Part of Speech Features and Hybrid Neural Network
摘要
Abstract
This paper proposes a recognition model for Chinese composition graceful sentences based on part-of-speech fea-tures and hybrid neural networks.Firstly,the terms in the sentence and the part-of-speech features corresponding to each term are vectorized and represented by fusion.Secondly,the convolutional neural network(CNN)with the introduction of attention mecha-nisms is used to obtain the local features of the fused word vectors.Bi-Gated recurrent unit(BiGRU)is used to serialize the fused word vectors to obtain global features.This paper fuses the two features and connects them to a classifier for graceful sentence recog-nition.Experiments show that the model based on part-of-speech features and hybrid neural network has the highest recognition ac-curacy for beautiful sentences,reaching 90.07%and F1 value reaching 83.13%.关键词
优美句/词性特征/卷积神经网络/注意力机制/双向门控循环神经网络Key words
elegant sentences/part-of-speech features/CNN/attention mechanisms/BiGRU分类
信息技术与安全科学引用本文复制引用
王芷恒,王芳,黄树成..基于词性特征和混合神经网络的中文作文优美句识别[J].计算机与数字工程,2025,53(4):995-1001,7.基金项目
国家自然科学基金项目"基于鲁棒表现建模的目标跟踪方法研究"(编号:61772244)资助. （编号:61772244）
