Naslov
Sentiment Analysis of Twitch Chats

Autori
Leonardo Čuljak

Vrsta, podvrsta i kategorija rada
Ocjenski radovi, diplomski rad, preddiplomski

Fakultet
Fakultet elektrotehnike i računarstva

Mjesto
Zagreb

Datum
01.07

Godina
2020

Stranica
38

Mentor
Bagić Babac, Marina

Ključne riječi
Twitch.TV ; sentiment analysis ; Natural Language Processing

Sažetak
Video gameplay has traditionally been social, with arcades providing an outlet for public play and consoles, allowing people to play together at home. Onlookers are not limited to being passive spectatorships as they are also provided the opportunity to engage directly with the player, with gameplay livestreaming being the newest iteration. Live streaming of video games has become dominated in most countries by the website Twitch.tv, where challenge concerns highly skilled players with minimal communication and exchange, conversely, prioritizes showmanship over gameplay. This thesis focuses on common communication characteristics of viewers who often send single messages and streamers who elaborate. Based on a large real dataset of livestream chats, this thesis uses sentiment analysis to explore and compare the findings of different communication theories regarding viewers and how they perceive streamers along the spectrum of different sentiments.

Izvorni jezik
Engleski

Znanstvena područja
Računarstvo

POVEZANOST RADA