Bibliographic citations
This is an automatically generated citacion. Modify it if you see fit
Ayvar, R., (2024). Detección de trampas en partidas de ajedrez en línea utilizando clasificadores de Machine Learning [Universidad de Lima]. https://hdl.handle.net/20.500.12724/21286
Ayvar, R., Detección de trampas en partidas de ajedrez en línea utilizando clasificadores de Machine Learning []. PE: Universidad de Lima; 2024. https://hdl.handle.net/20.500.12724/21286
@misc{renati/1438449,
title = "Detección de trampas en partidas de ajedrez en línea utilizando clasificadores de Machine Learning",
author = "Ayvar Rios, Rodrigo",
publisher = "Universidad de Lima",
year = "2024"
}
Title: Detección de trampas en partidas de ajedrez en línea utilizando clasificadores de Machine Learning
Authors(s): Ayvar Rios, Rodrigo
Advisor(s): Ramos Ponce, Oscar Efraín
OCDE field: https://purl.org/pe-repo/ocde/ford#2.02.04
Issue Date: 2024
Institution: Universidad de Lima
Abstract: El ajedrez en línea ha tenido un auge de popularidad desde el año 2020, con el contexto de la pandemia. Este auge trae consigo la problemática de las trampas en torneos de ajedrez en línea, las cuales comprometen su integridad y reputación, perjudicando tanto a jugadores como
organizadores. Paralelamente, se han utilizado diferentes técnicas de Machine Learning en ajedrez, en los ámbitos de clasificación y regresión, principalmente para la predicción de resultados. Sin embargo, existen pocos estudios sobre la detección de trampas en ajedrez, principalmente por la dificultad de acceso a partidas con trampas en línea. Por este motivo, el objetivo de esta investigación fue desarrollar cuatro modelos de detección de trampas con clasificadores de Machine Learning, utilizando partidas entre humanos y computadoras para caracterizar el comportamiento de jugadores haciendo trampas. Se construyó un dataset de entrenamiento a partir de las partidas de Free Internet Chess Server y un dataset de validación con partidas propias en el sitio web Chess.com. Además, se añadieron variables de análisis y de tiempo, sugeridas por investigaciones previas. El mejor modelo fue Random Forest, que obtuvo exactitudes del 96.534% con partidas del dataset de Free Internet Chess Server y del 96% con partidas en Chess.com. Esta investigación aporta al estado del arte, proporcionando modelos de detección de trampas en partidas de ajedrez con una mayor exactitud y alcance que estudios previos, siguiendo la definición de trampas hecha por la entidad reguladora de ajedrez en el mundo.
Online chess has had a rise in popularity since 2020, with the context of the pandemic. This rise brings with it the problem of cheating in online chess tournaments, which compromises their integrity and reputation, harming both players and organizers. In parallel, different Machine Learning techniques have been used in chess in the classification and regression fields. However, there are few studies on the detection of cheating in chess, mainly due to the difficulty of accessing cheating games online. For this reason, the objective of this research was to develop four cheat detection models with Machine Learning classifiers, using games between humans and computers to characterize the cheaters’ behavior. A dataset was built from the Free Internet Chess Server games and a validation dataset with own games on the Chess.com website. In addition, analysis and time variables were added, suggested by previous research. The best model was Random Forest, which obtained accuracies of 96.534% with games from the original dataset and 96% with games on Chess.com. This research contributes to the state of the art, providing models for detecting cheating in chess games with greater accuracy and reach than previous studies, following the definition of cheating made by the chess regulatory entity in the world.
Online chess has had a rise in popularity since 2020, with the context of the pandemic. This rise brings with it the problem of cheating in online chess tournaments, which compromises their integrity and reputation, harming both players and organizers. In parallel, different Machine Learning techniques have been used in chess in the classification and regression fields. However, there are few studies on the detection of cheating in chess, mainly due to the difficulty of accessing cheating games online. For this reason, the objective of this research was to develop four cheat detection models with Machine Learning classifiers, using games between humans and computers to characterize the cheaters’ behavior. A dataset was built from the Free Internet Chess Server games and a validation dataset with own games on the Chess.com website. In addition, analysis and time variables were added, suggested by previous research. The best model was Random Forest, which obtained accuracies of 96.534% with games from the original dataset and 96% with games on Chess.com. This research contributes to the state of the art, providing models for detecting cheating in chess games with greater accuracy and reach than previous studies, following the definition of cheating made by the chess regulatory entity in the world.
Link to repository: https://hdl.handle.net/20.500.12724/21286
Discipline: Ingeniería de Sistemas
Grade or title grantor: Universidad de Lima. Facultad de Ingeniería
Grade or title: Ingeniero de Sistemas
Juror: Quintana Cruz, Hernán Alejandro; Mayhua Quispe, Angela Gabriela; Quiroz Villalobos, Lennin Paul
Register date: 3-Oct-2024
This item is licensed under a Creative Commons License