A Text Mining Approach to Analyzing the Omnichannel Retail Business Performance of the KlikIndomaret App

The evolution of Web 2.0 technology has significantly influenced the use of Android applications, enabling users to provide feedback through reviews and star ratings. In managing omnichannel retail businesses, this user-generated content serves as a valuable source of information for performance evaluation and strategic management of both online and offline operations. Large-scale user review data is well-suited for analysis through text mining, particularly in sentiment analysis, when combined with topic and keyword filtering in the business domain. This study utilizes the RoBERTa Transformer model for the sentiment classification of user reviews. Among the 520 user reviews, 211 displayed good emotion, while 309 showed negative sentiment. By applying filtering processes to topics and keywords within the omnichannel retail business domain, the study identifies "economic value" and "delivery and CRM" as priority areas for improvement. This conclusion is drawn based on the significant disparity between positive and negative sentiments. As a result, management can formulate strategies to enhance the performance and user experience of the KlikIndomaret Android application.