基于特征权重的词向量文本表示模型  

Word vector text representation model based on feature weight

作　　者：蒋延杰 李云红[2] 苏雪平 张蕾涛 贾凯莉 陈锦妮[2] JIANG Yanjie;LI Yunhong;SU Xueping;ZHANG Leitao;JIA Kaili;CHEN Jinni(Beijing Administration for Code Allocation to Organization, Beijing 100010 ,China;School of Electronics and Information, Xi’an Polytechnic University, Xi’an 710048, China)

机构地区：[1]北京市组织机构代码管理中心,北京100010 [2]西安工程大学电子信息学院,陕西西安710048

出　　处：《西安工程大学学报》2022年第1期108-114,共7页Journal of Xi’an Polytechnic University

基　　金：国家自然科学基金(61902301);西安市科技局高校人才服务企业项目(2019217114GXRC007CG008-GXYD7.13)。

摘　　要：针对传统文本表示方法无法准确表达文本信息、稀疏维度高等问题,提出基于特征权重的词向量文本表示模型。通过Glove模型获得词向量,然后分别与TF-IDF、N-Gram模型相结合,分析考虑了文本的全局信息,解决了传统表示方法中稀疏维度高的问题,更好地捕捉了文本的语义和语序等局部信息,提高了文本特征表达能力。最后,通过20NewsGroup和5AbstractsGroup测试,分类准确率分别为85.93%、87.02%,验证了文本表示模型的有效性。Traditional text representation methods have inaccurate expression of text information and high sparse dimensions.A word vector text representation model based on feature weights was thus proposed.The word vector was obtained through the Glove model,and then combined with the TF-IDF and N-Gram models,which not only considers the global information of the text,but also solves the problem of high sparse dimensionality in traditional representation methods.This can better capture the local information such as text semantics and word order and improve the ability of text feature expression.Finally,through the 20NewsGroup and 5AbstractsGroup tests,the classification accuracy rates reached 85.93%and 87.02%,respectively,verifying the effectiveness of the text representation model.

关 键 词：TF-IDF模型 N-GRAM模型 Glove模型 文本表示 

分 类 号：TN912.34[电子电信—通信与信息系统]