基于评论数据挖掘的适老辅具及相关物流服务需求研究  

Research on Demand for Geriatric Aids and Related Logistics Services via Comment Mining

作　　者：李欣雨 罗鄂湘[1] 贾泽如 Xinyu Li;Exiang Luo;Zeru Jia(Business School,University of Shanghai for Science and Technology,Shanghai)

机构地区：[1]上海理工大学管理学院,上海

出　　处：《建模与仿真》2025年第2期748-757,共10页Modeling and Simulation

摘　　要：通过结合传统机器学习和深度学习算法的集成模型,对护理床用户的网络评论文本进行情感极性分类和需求挖掘,精确地识别用户对护理床产品潜在诉求。研究流程包括数据获取与预处理、模型构建与调参、以及用户需求分析三个主要阶段。首先,通过爬虫技术从京东平台获取护理床用户评论数据,并进行预处理,以构建训练和测试数据集。其次,利用Stacking算法集成多种传统机器学习模型和LSTM深度学习模型,在测试集上达到了90.34%的准确率。最终,结合LDA主题模型和7Rs物流服务理论,提取出影响用户体验的关键因素,为护理床产品的设计和改进提供了用户视角的洞见,也为其他康复辅具产品的用户需求分析提供了一种有效的分析框架。An ensemble model integrating traditional machine learning and deep learning algorithms was utilized to conduct sentiment polarity classification and demand mining on user reviews of nursing beds,accurately identifying potential user needs.The research process encompasses three main stages:data acquisition and preprocessing,model construction and tuning,and user demand analysis.Web crawling techniques were employed to collect user reviews of nursing beds from the JD.com platform,followed by preprocessing to construct training and testing datasets.The Stacking algorithm was then applied to integrate multiple traditional machine learning models and the LSTM deep learning model,achieving an accuracy rate of 90.34%on the test set.By combining the LDA topic model and the 7Rs logistics service theory,key factors affecting user experience were extracted,providing user-oriented insights for the design and improvement of nursing bed products.This approach also offers an effective analytical framework for user demand analysis of other rehabilitation assistive products.

关 键 词：集成算法 长短时记忆神经网络LSTM 评论分析 用户需求 

分 类 号：TP391.1[自动化与计算机技术—计算机应用技术]