首页 | 本学科首页   官方微博 | 高级检索  
     检索      

面向文本分类的深度置信网络特征提取方法研究
引用本文:易军凯,王超,李辉.面向文本分类的深度置信网络特征提取方法研究[J].北京化工大学学报(自然科学版),2018,45(3):90-94.
作者姓名:易军凯  王超  李辉
作者单位:北京化工大学 信息科学与技术学院,北京,100029;北京化工大学 信息科学与技术学院,北京,100029;北京化工大学 信息科学与技术学院,北京,100029
基金项目:NSFC-通用技术基础研究联合基金(U1636208)
摘    要:在对文本分类领域发展现状进行研究的基础上,提出了一种面向文本分类的深度置信网络特征提取方法,通过引入词向量模型和深度置信网络解决传统文本分类方法在文本表示及特征提取方面存在的语义缺失问题,实验结果表明,该方法在文本分类中有更高的准确率。

关 键 词:文本分类  深度学习  深度置信网络  词向量模型  特征提取
收稿时间:2017-08-21

A feature extraction method for text categorization based on a deep belief network (DBN)
YI JunKai,WANG Chao,LI Hui.A feature extraction method for text categorization based on a deep belief network (DBN)[J].Journal of Beijing University of Chemical Technology,2018,45(3):90-94.
Authors:YI JunKai  WANG Chao  LI Hui
Institution:College of Information Science and Technology, Beijing University of Chemical Technology, Beijing 100029, China
Abstract:Based on earlier research on the development of text categorization,in this paper,a new feature extraction method for text categorization is proposed.The method solves the semantic loss problem of text representation and feature extraction found in traditional text categorization methods by introducing word embedding and a deep belief network.Experiments show that the new method has higher accuracy than traditional methods in text categorization.
Keywords:text categorization                                                                                                                        deep learning                                                                                                                        deep belief network                                                                                                                        word embedding model                                                                                                                        feature extraction
本文献已被 CNKI 万方数据 等数据库收录!
点击此处可从《北京化工大学学报(自然科学版)》浏览原始摘要信息
点击此处可从《北京化工大学学报(自然科学版)》下载免费的PDF全文
设为首页 | 免责声明 | 关于勤云 | 加入收藏

Copyright©北京勤云科技发展有限公司  京ICP备09084417号