Research on the reputation of Chinese cigarette brands based on online comments

Background In China, cigarette brands vary, each with unique traits and consumer preferences. Analyzing online comments can truly show brand reputations and user experiences, supporting effective decision-making for cigarette market control. Methods This paper presents an online comment analysis method based on the Latent Dirichlet Allocation (LDA) topic-modeling technique. First, we built a tobacco-field sentiment lexicon. Then, we used the LDA model to identify cigarette brand influencing factors and extract high-frequency feature words. Finally, we gathered 152,383 comments of 25 cigarette brands and calculated the comprehensive evaluation value through sentiment analysis. Results Test data shows: 1) There are 4 key factors in tobacco brand evaluation: price, quality, service and packaging; 2) The comprehensive evaluation value of cigarette brands comparing with existing advanced models is an efficient reference of consumers’affective disposition and demonstrates greater specialized. Conclusion The LDA-based online comment analysis approach provides a basis for cigarette brand R&D and supply decisions, helping meet market and consumer needs better.