Primerjava in testiranje filmskih priporočilnih sistemov

BIHAR, TILEN (author), Zaletelj, Janez (mentor)

Diplomsko delo opisuje priporočilne sisteme s poudarkom na uporabniškem pogledu. Pri tem predstavimo tako uporabniško dojemanje priporočilnih sistemov kot tudi sistemsko dojemanje uporabnikov. V prvem delu predstavimo pomen in uporabnost priporočilnih sistemov. Poznavanje zgodovine razvoja priporočilnih sistemov nam omogoča razumevanje današnje pomembnosti. V nadaljevanju si ogledamo težave v delovanju in nadrobno spoznamo razliko med uporabniškimi vidiki in sistemskimi metrikami. V drugem delu primerjamo delovanje obstoječih filmskih priporočilnih sistemov Jinni, TasteKid, MovieLens, Criticker, Rotten Tomatoes, IMDb, Suggest Me Movie, A Good Movie To Watch, Flixter in Metacritic. Izvedemo tudi primerjavo in primer testiranja natančnosti na podlagi RMSE in MAE metrik.

• Language: Slovenian
• Keywords: priporočilni sistemi, uporabniški pogled, HCI, priporočanje filmov, testiranje natančnosti, RMSE, MAE
• Work type: Bachelor thesis/paper
• Organization: FE - Faculty of Electrical Engineering
• Year: 2015
• PID: 20.500.12556/RUL-72776
• Publication date in RUL: 30.09.2015

Secondary language

• Language: English
• Title: Comparison and evaluation of movie recommender systems

Abstract

The thesis describes the recommendation systems with an emphasis on user perspective. By this presents both user perception of recommendation systems as well as systemic perception of users. In the first part we present the importance and usefulness of recommendation systems. Knowing the history of the development of recommendation systems allow us to understand it's todays importance. Further we look at the problems in the operation in detail and realize the difference between the user perspective and system metrics. In the second part we compare the functioning of the existing film recommendation systems Jinni, TasteKid, MovieLens, Criticker, Rotten Tomatoes, IMDb, Suggest Me Movie A Good Movie To Watch, Flixter and Metacritic. We perform comparison and example of accuracy testing based on RMSE and MAE metrics.

• Keywords: recommendation systems, user perspective, HCI, recommending films, accuracy testing, RMSE, MAE