Application of Natural Language Processing to Extract Consumer Behaviors from Product Reviews

Abstract

Understanding consumer behavior is essential for effective marketing strategies. Market research often employs conventional qualitative analysis to uncover the motivations driving consumer purchases. However, a small sample size limits the ability to draw population-level conclusions, while analyzing large-scale data can be time-consuming. Moreover, qualitative research is prone to human error, subjectivity, and bias, leading to potentially misleading results. To tackle these issues, Natural Language Processing (NLP) is utilized to extract consumer behavior from large-scale product review datasets, representing typical purchasing behavior including complex, dissonance-reducing, habitual purchasing, and variety-seeking. The study focuses on luxury cars, smart TVs, bread, and soaps to investigate complex, dissonance-reducing, habitual, and variety-seeking purchasing behavior, respectively. Results are categorized using Perceived Value Scale, revealing that consumer intentions to purchase are influenced by perceived emotional, quality, and social value, respectively in complex purchasing behavior. Dissonance-reducing behavior was found to be primarily driven by perceived quality value, while habitual purchasing behavior was observed to be influenced by perceived economic, emotional, and quality value, respectively. Lastly, variety-seeking behavior was found to be guided by perceived emotional and quality value in purchase decisions, respectively. Therefore, NLP demonstrated an efficient, cost-effective, and unbiased approach to extracting consumer behaviors from large-scale product reviews.