An Analysis of the Sales and Consumer Preferences of E-cigarettes Based on Text Mining of Online Reviews

After viewing the literatures of text mining methods, we found that seldom have researchers studied the utility of text mining methods on massive online reviews. We draw more than 400000 reviews from an e-commerce platform and calculate the sentiment values and the frequency of feature words. Then we study the time dimension and geographical distribution of sales volume and sentiment values, as well as the geographical distribution of feature keywords. By doing this, we extract valuable information and discuss the consumer profiles of online users, coming out the conclusion that the sales volume isn't necessarily corresponding to sentiment strength and there exists differentiation in focus of people in different provinces.