This paper explores the application of dimensionality reduction techniques, specifically Singular Value Decomposition (SVD), in improving the performance of recommender systems. Recommender systems are used to suggest products to customers during live interactions, and they have become widely successful in e-commerce. However, they face challenges such as producing high-quality recommendations, making recommendations quickly for millions of customers and products, and achieving high coverage in the face of data sparsity. Collaborative filtering is a successful technique for recommender systems, but its performance degrades with the number of customers and products. The paper presents two experiments comparing the effectiveness of SVD and collaborative filtering in generating recommendations. The first experiment evaluates the ability of the systems to predict consumer preferences based on explicit ratings, while the second experiment assesses their ability to generate Top-N lists based on a real-life customer purchase database. The results suggest that SVD has the potential to meet many of the challenges of recommender systems under certain conditions. The paper discusses the limitations of existing collaborative filtering approaches, such as sparsity, scalability, and synonymy. SVD is proposed as a solution to these issues by reducing the dimensionality of the customer-product ratings matrix. This allows for better neighborhood formation and improved recommendation quality. The paper also describes the application of SVD in recommender systems, including the construction of SVD-based algorithms for prediction and top-N recommendations. The experiments show that SVD-based systems outperform collaborative filtering in terms of prediction accuracy and recommendation quality. The results indicate that SVD can provide better online performance and is more scalable for large-scale problems. The paper concludes that SVD has the potential to be a valuable tool in recommender systems, particularly for handling sparse data and improving recommendation quality. Future research is needed to further understand the effectiveness of SVD in different scenarios and to explore additional applications of dimensionality reduction in recommender systems.This paper explores the application of dimensionality reduction techniques, specifically Singular Value Decomposition (SVD), in improving the performance of recommender systems. Recommender systems are used to suggest products to customers during live interactions, and they have become widely successful in e-commerce. However, they face challenges such as producing high-quality recommendations, making recommendations quickly for millions of customers and products, and achieving high coverage in the face of data sparsity. Collaborative filtering is a successful technique for recommender systems, but its performance degrades with the number of customers and products. The paper presents two experiments comparing the effectiveness of SVD and collaborative filtering in generating recommendations. The first experiment evaluates the ability of the systems to predict consumer preferences based on explicit ratings, while the second experiment assesses their ability to generate Top-N lists based on a real-life customer purchase database. The results suggest that SVD has the potential to meet many of the challenges of recommender systems under certain conditions. The paper discusses the limitations of existing collaborative filtering approaches, such as sparsity, scalability, and synonymy. SVD is proposed as a solution to these issues by reducing the dimensionality of the customer-product ratings matrix. This allows for better neighborhood formation and improved recommendation quality. The paper also describes the application of SVD in recommender systems, including the construction of SVD-based algorithms for prediction and top-N recommendations. The experiments show that SVD-based systems outperform collaborative filtering in terms of prediction accuracy and recommendation quality. The results indicate that SVD can provide better online performance and is more scalable for large-scale problems. The paper concludes that SVD has the potential to be a valuable tool in recommender systems, particularly for handling sparse data and improving recommendation quality. Future research is needed to further understand the effectiveness of SVD in different scenarios and to explore additional applications of dimensionality reduction in recommender systems.