Digital Transformation and Administration Innovation 4 Serial Number 12 2026 04 01 A Sentiment Analysis–Based Recommender System for Online Retail Stores Using Customer Social Media Data A Sentiment Analysis–Based Recommender System for Online Retail Stores Using Customer Social Media Data 1 10 10.61838/dtai.239 EN Journal Article 2025 10 03 <p>This study aimed to design and empirically evaluate a sentiment analysis–based recommender system and to examine its effects on recommendation quality, customer trust, satisfaction, and repurchase intention in online retail environments. The study employed an applied mixed-methods design combining computational modeling with behavioral analysis. Social media data were collected from major online retail brands in Tehran, yielding approximately 1.8 million text records. A sample of 360 active online shoppers and 60 domain experts participated in system evaluation. Natural language processing techniques, including deep learning–based sentiment classification models, were used to extract emotional information from user-generated content. The recommender system integrated sentiment scores with transactional and behavioral data through a hybrid recommendation framework. System performance and behavioral effects were assessed using standard recommendation metrics and survey-based instruments. The sentiment-based recommender system significantly outperformed conventional collaborative filtering and content-based models in precision, recall, F1-score, normalized discounted cumulative gain, and click-through rate. Regression analysis revealed that recommendation quality had significant positive effects on customer trust (β = 0.62, p &lt; 0.001) and satisfaction (β = 0.58, p &lt; 0.001). Trust (β = 0.54, p &lt; 0.001) and satisfaction (β = 0.47, p &lt; 0.001) both significantly predicted repurchase intention. Post-implementation measures indicated significant increases in purchase intention, customer satisfaction, platform trust, and average order value (p &lt; 0.001). Integrating social media sentiment analysis into recommender systems substantially enhances system performance, customer engagement, and commercial outcomes, demonstrating the strategic value of emotionally intelligent personalization in online retail.</p> https://journaldtai.com/index.php/jdtai/article/download/239/230