• 首页
  • 期刊简介
  • 编委会
  • 投稿指南
  • 收录情况
  • 杂志订阅
  • 联系我们
引用本文:尤良辉,张华熊.基于丝绸行业电商评论的属性级情感分析[J].软件工程,2024,27(6):44-47.【点击复制】
【打印本页】   【下载PDF全文】   【查看/发表评论】  【下载PDF阅读器】  
←前一篇|后一篇→ 过刊浏览
分享到: 微信 更多
基于丝绸行业电商评论的属性级情感分析
尤良辉1, 张华熊2
(1.浙江理工大学信息科学与工程学院, 浙江 杭州 310000;
2.浙江理工大学计算机科学与技术学院, 浙江 杭州 310000)
202130504201@mails.zstu.edu.cn; zhxhz@zstu.edu.cn
摘 要: 随着移动互联网时代的快速发展,电商平台迅速崛起成为推动网络消费增长的一股新兴且强大的力量。为了有效利用海量的商品评论数据,文章基于京东商城丝绸商品的评论数据,使用词频统计对评论数据进行分析处理,构建属性-情感词词典,填充了评论中的隐性属性。利用Label Studio数据标注平台对评论数据进行属性-观点-情感的三元标注,经过标注后的数据集被应用于UIE(Unified Structure Generation for Universal Information Extraction)模型进行属性级情感抽取,并基于抽取的数据集对ERNIE(Enhanced Language Representation with Informative Entities)模型进行微调训练。实验结果表明,该方法在属性级情感分析中的准确率高达90%,填充隐性属性后,准确率提升至94%,表明该方法所得模型在属性级情感分析中有着不错的效果。
关键词: 电商评论;深度学习;UIE;属性级情感分析
中图分类号: TP391.9    文献标识码: A
Attribute-level Sentiment Analysis Based on E-commerceReviews in the Silk Industr
YOU Lianghui1, ZHANG Huaxiong2
(1.School of Information Science and Engineering, Zhejiang Sci-Tech University, Hangzhou 310000, China;
2.School of Computer Science and Technology, Zhejiang Sci-Tech University, Hangzhou 310000, China)
202130504201@mails.zstu.edu.cn; zhxhz@zstu.edu.cn
Abstract: With the rapid development of the mobile internet era, e-commerce platforms have quickly emerged as a new and powerful force driving the growth of online consumption. In order to effectively utilize massive product review data, this study is based on the review data of silk products on JD.com. Word frequency statistics are used to analyze and process the review data, construct an attribute-sentiment word dictionary, and fill in implicit attributes in the reviews. The Label Studio data labeling platform is used to perform triple labeling of attributes-opinions-sentiments on the review data. The labeled dataset is then applied to the UIE ( Unified Structure Generation for Universal Information Extraction) model for attribute-level sentiment extraction. The extracted dataset is used to fine-tune the ERNIE (Enhanced Language Representation with Informative Entities) model. Experimental results show that the proposed method achieves an accuracy rate of up to 90% in attribute-level sentiment analysis. After filling in implicit attributes, the accuracy rate increases to 94% , demonstrating that the model obtained by this method has a good effect in attribute-level sentiment analysis.
Keywords: E-commerce reviews; deep learning; UIE; attribute-level sentiment analysis


版权所有:软件工程杂志社
地址:辽宁省沈阳市浑南区新秀街2号 邮政编码:110179
电话:0411-84767887 传真:0411-84835089 Email:semagazine@neusoft.edu.cn
备案号:辽ICP备17007376号-1
技术支持:北京勤云科技发展有限公司

用微信扫一扫

用微信扫一扫