基于RoBERTa和Bi-GRU的文本情感分析研究  

Research on Sentiment Analysis Based on RoBERta and Bi-GRU

作　　者：罗嘉俊 付博宇 LUO Jiajun;FU Boyu(Xi'an University of Posts&Telecommunications,Xi'an Shaanxi 710100,China)

机构地区：[1]西安邮电大学,陕西西安安710100

出　　处：《信息与电脑》2023年第17期21-23,共3页Information & Computer

摘　　要：电商平台的用户评论是影响消费者决策的关键因素。在大量的评论中,如何正确地挖掘顾客的情绪,从而提高产品质量、服务质量、销售量,具有很大的商业价值。文章运用深度学习方法,对电子商务评论中的情绪倾向性进行了深入研究。采用了长短时记忆神经网络(Long Short-Term Memory,LSTM)的改进结构GRU+RoBERTa来提升情绪分析的精确性。实验结果证明,该模型应用效果较好。User comments on e-commerce platforms are a key factor affecting consumer decision-making In a large number of comments,how to correctly explore the emotions of customers,so as to improve product quality,service quality,sales volume,has great commercial value.This paper uses deep learning methods to conduct in-depth research on emotional tendencies in e-commerce reviews.The improved structure GRU+RoBERTa-WWM of the Long Short-Term Memory(LSTM)neural network was adopted to improve the accuracy of emotion analysis.The experimental results show that the model is effective.

关 键 词：情感分析 RoBERTa 双向门控循环网络 

分 类 号：TP391[自动化与计算机技术—计算机应用技术]