Usporedba kontekstno utemeljenih i kolaborativnih sustava za davanje preporuka te implementacija hibridnog sustava

master's thesis

• Title (english): Comparison of content-based and collaborative recommendation systems and implementation of a hybrid system
• Author: Dina Petrak
• Mentor: Marko Đurasević (mentor)
• Committee member: Marko Đurasević (predsjednik povjerenstva)
• Committee member: Domagoj Jakobović (član povjerenstva)
• Committee member: Nikolina Frid (član povjerenstva)
• Granter: University of Zagreb; Faculty of Electrical Engineering and Computing; Zagreb
• Defense date and country: 2025-02-19, Croatia

Abstract

U ovom radu opisan je rad sustava za davanje preporuka te je dana osnovna podjela sustava. Opisano je funkcioniranje sadržajnog sustava i sustava temeljenog na suradnji. Ti sustavi implementirani su korištenjem različitih metrika te su uspoređene performanse. Analizirane su prednosti i nedostatci svakog pristupa. Posebni interes je usmjeren na problem hladnog starta te funkcioniranje sustava u situacijama kada nedostaje podataka. Ispitane su i implementirane alternativne verzije sustava koje se... More dobro nose s manjkom podataka. Implementiran je hibridni sustav koji težinski kombinira sadržajne i suradničke preporuke, te koristi demografske preporuke u situaciji novih korisnika. Hibridni sustav testiran je evaluacijom izvan vremenskog okvira. Osim toga, provedena je analiza zadovoljstva korisnika s preporukama, iz koje su izvedeni zaključci o kvalitetama i nedostatcima sustava koji su kombinirani te hibridnog sustava. Less

Abstract (english)

This paper describes how recommendation systems work and depicts the main classification. The operation of content-based recommendation and collaborative filtering is explained. Both systems were implemented using different metrics and their performances were compared. The advantages and disadvantages of each approach were analyzed. The main focus was directed to the cold start problem and the functioning of the systems in situations of data shortage. Alternative versions of the recommendation... More system that cope well with lack of data have been tested and implemented. A hybrid system has been implemented that combines content and collaborative recommendations by a weighted approach and uses demographic recommendations in the situation of new users. The hybrid system was tested with out-of-time evaluation. In addition, an analysis of user satisfaction with recommendations was carried out, from which conclusions were drawn about the qualities and shortcomings of the combined systems and the hybrid system. Less

• Keywords: Sustav za filmske preporuke; preporuke temeljene na sadržaju; kolaborativno filtriranje; problem hladnog starta; demografske preporuke; hibridni sustav preporuka
• Keywords (english): Movie recommendation system; content-based recommendation; collaborative filtering; cold start problem; demographic recommendation; hybrid recommendation system
• Language: croatian
• URN:NBN: urn:nbn:hr:168:853755
• Study programme: Title: Computing; Study programme type: university; Study level: graduate; Academic / professional title: magistar/magistra inženjer/inženjerka računarstva (magistar/magistra inženjer/inženjerka računarstva)
• Type of resource: Text
• File origin: Born digital
• Access conditions: Open access
• Created on: 2025-03-12 11:15:56