Banca de DEFESA: ISAAC NEWTON DA SILVA BESERRA

Uma banca de DEFESA de MESTRADO foi cadastrada pelo programa.
DISCENTE : ISAAC NEWTON DA SILVA BESERRA
DATA : 26/06/2017
HORA: 10:30
LOCAL: Auditório I - DIMAp
TÍTULO:

Acquisition and analysis of the first keystroke dynamics biometrics database for user verification in the online collaborative game League of Legends.

PALAVRAS-CHAVES:

Online Game, Keystroke Dynamics, League of Legends, Biometrics

PÁGINAS: 50
GRANDE ÁREA: Ciências Exatas e da Terra
ÁREA: Ciência da Computação
SUBÁREA: Metodologia e Técnicas da Computação
ESPECIALIDADE: Processamento Gráfico (Graphics)
RESUMO:

The popularity of computer games has grown exponentially in the last few years,
reaching a point where they stopped being just a child's play and became actual sports.
Being so, many players invest lots of time and money to improve and become the best. As
a result, many online games companies protect users' accounts in a variety of ways, such
as secondary passwords, e-mail conrmation when accessed on other devices and mobile
text messages. However, none of these techniques apply when it comes to account sharing.
In the competitive scenario, players are divided by their level of skill, which is obtained
from their achievements and victories; thus, when a player shares his/her account, s/he is
classied in a level which does not correspond to his/her actual skill, causing an imbalance
in matches. The game League of Legends greatly suers with this practice, which is
popularly known as Elo Job, which is forbidden by the game company itself and, when
discovered, causes the player to be permanently banned from the game. As the game
uses the keyboard keystroke dynamics for most of its actions, a continuous verication
during the game would be ideal, as it could potentially identify whether the player is
really the owner of the account. As a result, the system could penalise players who share
their accounts. For this work, a keystroke-based biometrics database was populated with
data collected from real players. The data were analyzed and tested with several classiers,
obtaining a hit rate of 65.90%, which is not enough to make a good identication. However,
the combination of the features of the keystroke dynamics with the mouse dynamics
showed much better results, reaching a promising hit rate of 90.0%.

MEMBROS DA BANCA:
Interno - 2177445 - BRUNO MOTTA DE CARVALHO
Interno - 2524467 - MARJORY CRISTIANY DA COSTA ABREU
Externo ao Programa - 2978747 - CHARLES ANDRYE GALVAO MADEIRA
Externo à Instituição - CARLOS NASCIMENTO SILLA JUNIOR - PUCPR