文献类型：期刊文献

中文题名：基于深度学习与《中国图书馆分类法》的文献自动分类系统研究

英文题名：Research on Automatic Literature Classification System Based on Deep Learning and Chinese Library Classification

作者：孔洁[1]

机构：[1]中国劳动关系学院图书馆,北京100048

年份：2021

期号：5

起止页码：51

中文期刊名：新世纪图书馆

外文期刊名：New Century Library

收录：NSSD、CSSCI_E2021_2022、CSSCI

语种：中文

中文关键词：智能图书馆;深度学习;卷积神经网络;文献分类

外文关键词：Intelligent library;Deep learning;Convolutional neural network;Literature classification

中文摘要：为了弥补传统文献分类方法的不足,满足信息时代下激增的文献分类需求,文章提出了一种文献自动分类算法,结合NLPIR分词系统与Skim-gram词向量模型提取文献的特征向量矩阵,并在此基础上结合卷积神经网络对文献的中图法分类号进行预测。实验结果显示,文章提出模型的基本大类准确率为97.66%,二级分类准确率为95.12%,详细分类的准确率为92.42%。结果证明,结合特征词向量预处理与卷积神经网络能够有效提升文献分类精度,这为实现智能图书分类提供了新的思路。

外文摘要：In order to make up for the shortage of traditional literature classification methods and meet the demand of literature classification in the information age,this paper proposes an automatic literature classification algorithm,which combines NLPIR word segmentation system and Skim-gram word vector model to extract the eigenvector matrix of the literature,and on this basis,combine with the convolution neural network to predict the classification number of the literature.The experimental results show that the accuracy of the model proposed in this paper is 97.66%for basic categories,95.12%for secondary classification,and 92.42%for detailed classification.The results show that the combination of feature vector preprocessing and convolutional neural network can effectively improve the classification accuracy of literature,which provides a new idea for the realization of intelligent book classification.