Real-World Evaluation: Hybrid Recommender System and User Engagement

Recommender systems are essential for social media platforms, personalizing content suggestions, and driving user engagement. This study investigates the impact of a hybrid recommender system on user behavior within a real-world social media platform. The system combines content-based filtering and collaborative filtering techniques to deliver personalized recommendations to users. User engagement is evaluated over six months using various metrics beyond clicks, including likes, plays, stream time, comments, shares, and returning users. Analysis revealed significant improvements in user engagement following system implementation. Users displayed a 12.815% increase in likes, indicating heightened interest in recommended content. Plays surged by 124.159%, suggesting a solid alignment between recommendations and user preferences. Stream time also saw a substantial rise of 117.349%, further highlighting deeper user engagement with the recommended content. Positive trends were also observed in engagement depth, with increases of 5.589% in comments and 59.825% in shares, signifying user satisfaction and a willingness to interact with and share the discovered content. Notably, the number of returning users witnessed a remarkable growth of 287.704%, demonstrating the system's effectiveness in fostering user retention. These findings contribute to a deeper understanding of hybrid recommender systems and their potential to enhance user experience in social media environments. The study highlights how these systems can personalize content discovery, increase user satisfaction, and ultimately strengthen platform loyalty.