Your browser does not allow JavaScript!
JavaScript is necessary for the proper functioning of this website. Please enable JavaScript or use a modern browser.
Open Science Slovenia
Open Science
DiKUL
slv
|
eng
Search
Browse
New in RUL
About RUL
In numbers
Help
Sign in
Priporočilni sistem za predlaganje iger v spletni igralnici
ID
Bosil, Tara Patricija
(
Author
),
ID
Demšar, Jure
(
Mentor
)
More about this mentor...
PDF - Presentation file,
Download
(1,29 MB)
MD5: 29B1D221E560225DD60B07FFE201CC40
Image galllery
Abstract
Spletne igralnice in stavnice predstavljajo vse bolj priljubljen način zabave. Ker je konkurenca na trgu zelo velika, je zadovoljstvo stranke ključnega pomena. Pomembno vlogo pri tem ima personalizacija celotne izkušnje, ki jo lahko izboljšamo z uporabo priporočilnih sistemov. Cilj magistrske naloge je razvoj in evalvacija priporočilnih sistemov, ki igralcu spletne igralnice priporoča igre, ki ga utegnejo zanimati. Pri gradnji priporočilnih sistemov smo uporabili matrično faktorizacijo, faktorizacijske metode in ekstremno gradientno spodbujanje. Za potrebe evalvacije smo implementirali tudi nekaj naivnih algoritmov. Uporabili pa smo tudi hibridni način priporočanja, kjer kombiniramo naprednejše algoritme z naivnimi. Algoritme smo ovrednotili s pomočjo prečne validacije na podlagi historičnih podatkov. Naši rezultati nakazujejo, da najboljše napovedi dobimo s kombinacijo algoritma XGBoost in naivnega algoritma PMplays. A tega ne moremo zagotovo trditi, saj so metrike najboljših algoritmov znotraj napake. Da bi to potrdili, bi potrebovali dodatne eksperimente. Opazili smo tudi, da so pri gradnji algoritmov zelo pomembne značilke, ki opisujejo temo, čas in lokacijo igre. Rezultati naloge bodo podjetju služili kot izhodišče za gradnjo priporočilnega sistema v produkcijskem okolju. Pričakujemo, da bo implementiran priporočilni sistem pozitivno vplival na zadovoljstvo strank ter s tem tudi na uspešnost podjetja.
Language:
Slovenian
Keywords:
priporočilni sistem
,
faktorizacijske metode
,
matrična faktorizacija
,
ekstremno gradientno spodbujevanje
,
spletni kazinoji
Work type:
Master's thesis/paper
Typology:
2.09 - Master's Thesis
Organization:
FRI - Faculty of Computer and Information Science
Year:
2021
PID:
20.500.12556/RUL-133068
COBISS.SI-ID:
87405315
Publication date in RUL:
10.11.2021
Views:
1688
Downloads:
275
Metadata:
Cite this work
Plain text
BibTeX
EndNote XML
EndNote/Refer
RIS
ABNT
ACM Ref
AMA
APA
Chicago 17th Author-Date
Harvard
IEEE
ISO 690
MLA
Vancouver
:
Copy citation
Share:
Secondary language
Language:
English
Title:
Recommender system for games in an online casino
Abstract:
Online casinos and betting sites are becoming an increasingly popular way of having fun. Due to a fierce competition on the market, clients satisfaction is of vital importance. Personalization of the whole experience plays a key role in increasing customer satisfaction and recommender systems play a vital part of it. The goal of this master thesis is to develop and evaluate recommender systems that would recommend new and interesting games. In the construction of the recommender systems we used matrix factorization, factorization machines and extreme gradient boosting. For a better evaluation we also implemented some naive algorithms. On top of that we used a hybrid way of recommendation, where we combined more advanced algorithms with the naive ones. We evaluate our algorithms using cross validation on historical data. Our results indicate that a combination of the XGBoost algorithm and the naive PMplays algorithm produces the best prediction results. Additionally we noticed that game features that describe the theme, location and time of a game play an extremely important role in the building of our recommender systems. The results of this thesis will serve as a starting point in building a recommender system that will be used in production. We expect that the implemented recommender system will have a positive impact on customer satisfaction and thus on the overall success of the company.
Keywords:
recommender system
,
factorization machines
,
matrix factorization
,
extreme gradient boosting
,
online casino
Similar documents
Similar works from RUL:
Similar works from other Slovenian collections:
Back