基于机器学习的电商平台客户投诉精准定位研究  

Research on Precise Positioning of Customer Complaints on e-Commerce Platforms Based on Machine Learning

作　　者：景奕昕 JING Yixin(Wuhan IPASON Technology Co.,Ltd.,Wuhan 430000,China)

机构地区：[1]武汉攀升鼎承科技有限公司,湖北武汉430000

出　　处：《数字通信世界》2025年第1期22-24,27,共4页Digital Communication World

摘　　要：针对当前电商平台中客户投诉处理效率低下、解决周期长且准确率不高等问题,本文提出一种基于机器学习的客户投诉精准定位方法;针对收集到的客户投诉文本进行预处理,以确保后续分析的有效性;采用BERT等深度学习技术来提取投诉文本中的关键特征,以反映投诉的主要内容和情感倾向;基于此特征集,构建一个卷积神经网络模型,自动识别和分类客户投诉的不同类型及其优先级;最后,通过对比实验验证了所提方法的有效性和优越性。本方法的整体平均F1分数为0.96,预测标签与真实标签之间的差异程度为0.05,紧急投诉响应时间预测误差仅为0.45 h,为电商平台提供了一套高效、可靠的客户投诉管理方案。In response to the low efficiency,long resolution cycle,and low accuracy of customer complaint handling in current e-commerce platforms,this article proposes a machine learning based method for accurate customer complaint localization;Preprocess the collected customer complaint texts to ensure the effectiveness of subsequent analysis;Using deep learning techniques such as BERT to extract key features from complaint texts to reflect the main content and emotional tendencies of the complaint;Based on this feature set,construct a convolutional neural network model to automatically identify and classify different types of customer complaints and their priorities.Finally,the article validated the effectiveness and superiority of the proposed method through comparative experiments.The results showed that the overall average F1 score of this design was 0.96,the degree of difference between predicted labels and real labels was 0.05,and the prediction error of emergency complaint response time was only 0.45 hours,providing an efficient and reliable customer complaint management solution for e-commerce platforms.

关 键 词：机器学习 电商平台 投诉定位 

分 类 号：F713.36[经济管理—产业经济] TP393.4[自动化与计算机技术—计算机应用技术]