科技创新与应用2024，Vol.14Issue(2)：54-58,5.DOI:10.19981/j.CN23-1581/G3.2024.02.013
基于BiGRU与胶囊网络的中文新闻标题文本分类
黄玉兰 1刘瑞安 1胡昕 1任超 1徐宇辉1
作者信息
摘要
Abstract
In order to make up for the importance and dependence of the words in the text sequence that can not be recognized by the traditional capsule network,this paper proposes a capsule text classification model-BMCapsNet model,which combines BiGRU and multi-head attention mechanism.First of all,the model uses BiGRU and long attention mechanism to obtain the global features of the text,and then uses the capsule network to extract deeper semantic information and classifies the text through capsule prediction.It is applied to the text classification task of Chinese news headlines,and the effectiveness of the model is proved in the THUCNews headline data set and Jinri Toutiao news title data set.关键词
BiGRU/BMCapsNet/多头注意力机制/中文新闻标题/文本分类Key words
BiGRU/BMCapsNet/multi-head attention mechanism/Chinese news headline/text classification分类
信息技术与安全科学引用本文复制引用
黄玉兰,刘瑞安,胡昕,任超,徐宇辉..基于BiGRU与胶囊网络的中文新闻标题文本分类[J].科技创新与应用,2024,14(2):54-58,5.
