FAIRNESS-AWARE RECOMMENDER SYSTEMS IN PRACTICE: A SCOPING REVIEW OF MODELS, EVALUATION PROTOCOLS, AND OPEN CHALLENGES

The recommender systems have emerged as a vital part of digital information platforms, such as those in e-commerce, education, healthcare, and entertainment. However, growing concerns about algorithmic bias, inequality, and the lack of transparency in recommendation systems have increased the need for fairness-aware recommender systems (FARS). Although there has been considerable research development, current studies are scattered across fairness definitions, modelling methods, evaluation methods, and application areas. This scoping review aims to outline the current literature on fairness-aware recommender systems, summarize existing modelling and evaluation methods, present practical applications, and identify research gaps and future directions. A scoping review was conducted using Google Scholar, ResearchGate, ScienceDirect, and Springer, with searches performed up to April 2026 without publication year or geographical restrictions. A total of 1203 records were identified, 557 were screened, 53 underwent full-text review, and 20 studies were included in the final synthesis. A thematic synthesis was conducted to examine fairness concepts, recommendation models, evaluation methods, application domains, and research challenges. The extracted studies were summarized into four themes: fairness concepts and bias sources, fairness-aware recommendation models, evaluation and fairness measurement, and practical applications. The findings indicate that fairness-aware recommender systems increasingly integrate fairness objectives with recommendation performance, while highlighting persistent gaps in evaluation standardization, real-world deployment, scalability, interpretability, and cross-domain validation. Overall, the review demonstrates that fairness-aware recommender systems are fundamental to the development of responsible and trustworthy AI, while emphasizing the need for standardized evaluation frameworks, explainable models, and stronger governance mechanisms to support their effective deployment.