Text sentiment analysis of online cigarette reviews
YANG Chunxiao1, ZHANG Hexin2, HUANG Jiawen3, WAN Jiangping3
1 Guangdong Twenty Commercial Chain Co., Ltd., Guangzhou 510610, China; 2 Guangdong Provincial Tobacco Company, Guangzhou 510623, China; 3 Business Administration College, South China University of Technology, Guangzhou 510640, China
Abstract [Background and objective] Cigarette online reviews can truly reflect the reputation of cigarettes, and help cigarette companies understand the consumer experience of cigarettes and the dynamics of cigarette reputation. In order to extract effective consumer experience information from massive online comments on cigarette consumption, this paper conducts a text sentiment analysis study of online review of cigarette.[Method] This paper firstly constructs a cigarette online commentary sentiment dictionary, and then establishes a sentiment analysis model for cigarette evaluation text. Two cigarette products with more comments are chosen as the research objects from the Yanyue network. The online reviews of the two products from 2008 to 2018 are collected using Python universal web crawler, and then cleaned to obtain effective online comment data. The comment data are input into the cigarette evaluation text sentiment analysis model, and the emotion index is calculated according to the dimensions of product, time, region through comparison of emotional indexes, we can gain insight into the emotional expression in the online comments of cigarettes, and give corresponding countermeasures and suggestions.[Conclusion] The sentiment index can provide new insights for cigarette management evaluation methods, dynamically monitor consumers' emotional changes, and timely grasp consumers' sentiment trends in the tobacco market environment.