Movie Recommender System Based on Collaborative Filtering

Ovaj se rad ukratko osvrće na motivaciju iza sustava preporučivanja te njihove najvažnije vrste. Objašnjeni su principi rada sustava preporučivanja zasnovanih na sadržaju, kao i onih zasnovanih na suradničkom filtriranju uporabom poznatijih primjera. Suradničko filtriranje pojašnjeno je kroz podjelu na pristupe zasnovane na modelu i one zasnovane na memoriji, s naglaskom na potonjem. Nadalje, spomenute su najčešće mjere točnosti koje se koriste za evaluaciju sustava preporučivanja. Osim toga, sadržan je postupak izgradnje sustava preporučivanja filmova uporabom suradničkog filtriranja zasnovanog na memoriji te njegova evaluacija.This paper briefly covers the motivation behind recommender systems and their most important types. It explains basic principles of content-based recommender systems, as well as those based on collaborative filtering using well-known examples. Collaborative filtering is clarified using division amongst model-based and memory-based approaches, with the emphasis on the latter. Furthermore, it mentions most common accuracy measures used to evaluate recommender systems. In addition, it documents the process of building a movie recommender system using memory based collaborative filtering, as well as its evaluation