基于评论大数据的手机产品改进  

Mobil phone product improvement based on big data of comment

作　　者：杨程[1] 谭昆 俞春阳 YANG Cheng;TAN Kun;YU Chunyang(Department of Industrial Design, Zhejiang University City College, Hangzhou 310015, China;College of Computer Science and Technology, Zhejiang University, Hangzhou 310027, China)

机构地区：[1]浙江大学城市学院工业设计系,浙江杭州310015 [2]浙江大学计算机科学与技术学院,浙江杭州310027

出　　处：《计算机集成制造系统》2020年第11期3074-3083,共10页Computer Integrated Manufacturing Systems

基　　金：国家自然科学基金资助项目(61672451);浙江省自然科学基金资助项目(LY18E050014);教育部人文社会科学研究规划基金资助项目(17YJAZH103)。

摘　　要：为应对高频产品迭代,提出一种基于评论大数据的产品设计改进研究方法。以手机为研究对象爬取大量在线评论,借助隐含狄利克雷主题模型确定用户关注的产品属性及其对应的属性词,并对评论的有用性进行分析;通过情感分析计算用户对手机各产品属性的关注度和满意度等指标,从而建立评价指标体系,并找到手机亟需改进的产品属性;通过观点挖掘找到用户不满意的点,进而确定手机的改进策略。利用多元线性回归分析建立手机评价指标与评论差评率的线性回归模型,通过采用该模型较准确地预测手机评论的差评率,验证了所提方法的有效性。该方法将挖掘和情感分析结合,可快速为产品设计改进提供决策依据。In response to the high frequency product iterations today,a research method on improvement of product design based on big data of comment was proposed.Taking mobile phone for research object,a large number of online comments were crawled.The product attributes concerned by users and their corresponding attribute words were determined by Latent Dirichlet Allocation,and comment helpfulness analysis was conducted.Through sentiment analysis,the relevant indicators such as the user's attention and satisfaction to each product attribute of phones were calculated to establish evaluation indicator system,and the product attributes of phones that needed to be improved were found.The user's dissatisfaction was found by opinion mining to determine the phone improvement strategy.A linear regression model between the phone evaluation indicators and the rate of negative comments was established through multiple linear regression analysis.The model could be used to accurately predict the rate of negative comments,which verified the effectiveness of the proposed method.By combining text mining and sentiment analysis,the method could quickly provide decision-making basis for improvement of product design.

关 键 词：大数据 文本挖掘 情感分析 产品设计改进 多元线性回归 手机 

分 类 号：TP391[自动化与计算机技术—计算机应用技术] TB472[自动化与计算机技术—计算机科学与技术]