Sustav preporučivanja filmova zasnovan na sadržaju

undergraduate thesis

• Title (english): Content-based Movie Recommendation System
• Author: Bartol Vinković
• Mentor: Klemo Vladimir (mentor)
• Committee member: Klemo Vladimir (predsjednik povjerenstva)
• Committee member: Adrian Satja Kurdija (član povjerenstva)
• Committee member: Lucija Šikić (član povjerenstva)
• Granter: University of Zagreb; Faculty of Electrical Engineering and Computing; Zagreb
• Defense date and country: 2021-06-29, Croatia
• Scientific / art field, discipline and subdiscipline: TECHNICAL SCIENCES; Computing

Abstract

Sustav preporučivanja na temelju sadržaja koristi mjere udaljenosti i sličnosti nad atributa objekta da izračuna sveukupnu sličnost dvaju objekata. Pojmovi sličnosti i udaljenosti su međusobno povezani u nekim slučajevima s izravnom relacijom. Kako bismo mogli napraviti takav sustav potrebno je pronaći prikladan skup podataka i filtrirati korisne podatke i urediti u smislenu cjelinu. Kako bismo bili sigurni da je naš sustav dobro izgrađen moramo provesti pokuse. Najdostupniji pokusi u offline pokusi za koje je samo potrebno imati korisničke podatke prema kojima možemo modelirati korisnika. Najčešći način evaluacije sustava preporučivanja je usporedba s drugim sustavom. Usporedbe radimo testove kojima povlačimo zaključke koji sustav ima bolje performanse, te statistički obradimo rezultate kako bismo bili sigurni da su rezultati pouzdani. Prilikom offline pokusa moramo modelirati korisnika, ali pripaziti da naš model ne odstupa previše od pravog ponašanja korisnika.

Abstract (english)

Content-based recommendation systems give recommendations using distance and similarity measures to determine the similarity of attributes of two objects and the total similarity. Similarity and distance are related terms, and in some cases have a direct relation with each other. In order to make such a system we need an appropriate dataset which we need to filter and organize. We need to conduct experiments to ensure that our system is working properly. The most available experiments are offline experiments for which we only need user-data that we use to model the users. The most common way of evaluating a recommender system is to compare it to another. The comparison is made with tests which determine which system better predicts user behavior. We conduct statistical analysis to determine that our results are repeatable. During the evaluation we need to make sure that our user model doesn't stray too much from real user behavior.

• Keywords: preporučivanje; sličnost; obrada velikih skupova podataka; sadržaj
• Keywords (english): recommendation; similarity; big data; content
• Language: croatian
• URN:NBN: urn:nbn:hr:168:889643
• Study programme: Title: Computing; Study programme type: university; Study level: undergraduate; Academic / professional title: sveučilišni/a prvostupnik/ prvostupnica (baccalaureus/baccalaurea) inženjer/inženjerka računarstva (sveučilišni/a prvostupnik/ prvostupnica (baccalaureus/baccalaurea) inženjer/inženjerka računarstva)
• Type of resource: Text
• File origin: Born digital
• Access conditions: Closed access
• Created on: 2023-01-14 10:12:15