Research, Development and Application of Algorithms in Recommender Systems

This master thesis is dedicated to the topic of the research, development, and application of algorithms in recommender systems. The author has done a comprehensive study that included a literature review and a hands-on case study. The relevance of this research topic is caused by the wide range of areas of recommender systems’ application across various domains such as e-commerce and personalized content delivery. Improving the effectiveness of recommender systems usually positively influences user experience, clients’ metrics and over-all financial results. The goal of this work is to develop a deep understanding of recommender systems, including their history and application areas, exploration of the classic and new algorithms employed in these systems and evaluation metrics used to assess its performance. For the case study the author used the ml-100k dataset provided by the MovieLens company, which contains movie ratings given by users. The performance of two Collaborative Filtering algorithms, specifically K-Nearest Neighbors (KNN) and Singular Value Decomposition (SVD), is assessed using this dataset. The results of the case study are analyzed and interpreted in terms of the effectiveness and relevance of these algorithms application in the given context.