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

Abstract: 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.

Key words: big data, text mining, sentiment analysis, improvement of product design, multiple linear regression, mobil phone