Text sentiment analysis from commodity evaluation data: case study of intelligent refrigerator reviews

Abstract

With the development of technology and the popularity of cell phones, online shopping behavior has penetrated into all aspects of people's lives, and more and more people have conducted online shopping. Product reviews have become an important reference for consumers to choose when shopping, and an important way for merchants to collect consumers' opinions. Therefore, it has become an important issue for consumers to get useful information in many reviews, and merchants to extract consumers' concerns and complaints about products through reviews, so as to design products that make consumers more satisfied and stand out in the increasingly fierce competition. In this paper, take the reviews of five different refrigerators in Jingdong Mall as an example, and use web crawler technology and sentiment analysis technology, and mine the review data for analysis. The specific steps are as follows: firstly, the crawled review data is pre-processed, and the sentimental tendency of text reviews is obtained using a dictionary-based sentiment analysis method. Based on this, further build the LDA (Latent Dirichlet Allocation) topic model to extract the potential topic words in the comments and dig out the deeper keywords in the comments. This will show the consumers' suggestions and concerns about the products more clearly, so that merchants can make corresponding adjustments and improve their competitiveness