摘要
本文以中国知网数据库收录的877篇文献和国外Web of Science核心合集数据库收录的423篇文献作为研究对象,运用CiteSpace软件构建了研究热点和研究前沿主题的可视化知识图谱,对国内外基于神经网络的网络舆情情感分析研究现状进行分析,并且对其差异进行对比。研究发现,国内外对该领域日趋关注但各有侧重,国内作者及机构存在相互合作较少,学术交流不足,研究主题较集中等问题。国内研究热点为情感分类、词向量、深度学习等;研究前沿趋势主要为特征融合表示方法、预训练模型的探索和词嵌入层的设计,为研究发展提供理论导向。
Taking 877 documents included in the domestic HowNet database and 423 documents included in the foreign web of science core collection database as the research object,this paper constructs a visual knowledge map of research hotspots and cutting-edge topics by using CiteSpace software,analyzes the research status of network public opinion and emotion analysis based on neural network at home and abroad,and compares their differences.It is found that more and more attention has been paid to this field at home and abroad,but each has its own emphasis.Domestic authors and institutions have problems such as less mutual cooperation,insufficient academic exchanges,and more concentrated research topics.Domestic research focuses on emotion classification,word vector,deep learning and so on;The research frontier trends mainly include the representation method of feature fusion,the exploration of pre training model and the design of word embedding layer,which provides a theoretical guidance for research and development.
作者
谭坤彦
杨孔雨
TAN Kunyan;YANG Kongyu(School of Information Management,Beijing Information Science and Technology University,Beijing 100192,China)
出处
《智能计算机与应用》
2022年第8期33-42,共10页
Intelligent Computer and Applications
基金
北京市社会科学基金重点项目(15ZHA004)
北京信息科技大学“促进高校分类发展-学科建设与研究生教育”项目资助。
