Discrete Emotions Response after Purchase Cosmetics of Millennial Consumers: Evaluation of Satisfaction/Normal/Dissatisfaction Attributes Using Text Mining Techniques

Purpose: This study aimed to analyze the similarities and differences between each area using wordcloud analysis and semantic network analysis, which are text mining techniques. Further, confirmatory factor analysis will be conducted by crawling for word-of-mouth information on attribute reviews as satisfied, normal, or dissatisfied after purchases that are subjectively given by millennial generations.Methods: The R program version 4.1.2 was used as a big data collection and analysis tool, and text mining analysis was performed through preprocessing and stopword processing on the collected data. Further, using LISREL 8.80 we conducted confirmatory factor analysis on these results.Results: Wordcloud analysis revealed that the terms “skin,” “products,” and “skin” ranked first in the evaluation area of “satisfied,” “normal,” and “dissatisfied,” respectively. Additionally, using confirmatory factor analysis, the correlation between the three latent variables of satisfaction, normal, and dissatisfaction was differentiated.Conclusion: The similarities and differences between the domains obtained through wordcloud and semantic network analyses and derived by classifying individual emotional responses of millennial consumers in social media into satisfied, normal, and dissatisfied domains are considered very meaningful. The keywords derived with high centrality in the semantic network for each domain is then refined and introduced as an observation variable for confirmatory factor analysis in accordance with the purpose of the study; this is helpful in research development for causal analysis in the future.