Thesis Estudio exploratorio del crimen organizado en la prensa en Chile desde 2014 aplicando técnicas de PLN + ML
Loading...
Date
2023
Authors
Journal Title
Journal ISSN
Volume Title
Program
Ingeniería Civil Informática
Departament
Campus
Campus Santiago San Joaquín
Abstract
Las organizaciones criminales dedicadas al narcotráfico son grupos delictuales que constantemente buscan la oportunidad para cometer sus delitos y es de conocimiento ciudadano que día tras día acontecen estos hechos que son informados tanto en periódicos populares como rurales. La cantidad de noticias policiales es tan grande que evaluar el comportamiento de estas organizaciones a través de los años puede ser muy complejo. Este documento propone un análisis temático de una gran cantidad de artículos periodísticos policiales desde 2014 utilizando algoritmos de aprendizaje automático no supervisado. El algoritmo de agrupamiento k-means facilita la observación de aquellos grupos de delitos correspondientes a crimen organizado y el algoritmo Latent Dirichlet Allocation permite obtener temas relevantes como la violencia aplicada a través del uso de armas, las relaciones familiares dentro de estas organizaciones y el cultivo de Cannabis. Además, es posible detectar conceptos no tendenciales relacionados a drogas sintéticas donde en el último tiempo destacan el uso de 2CB, ketamina y el fentanilo que actualmente atraviesa una crisis en Chile.
Criminal organizations dedicated to drug trafficking are criminal groups that constantly look for the opportunity to commit their crimes and it is common knowledge that these events occur day after day and are reported in both popular and rural newspapers. The amount of police news is so large that evaluating the behavior of these organizations over the years can be very complex. This paper proposes a thematic analysis of a large number of police news articles since 2014 using unsupervised machine learning algorithms. The k-means clustering algorithm facilitates the observation of those groups of crimes corresponding to organized crime and the Latent Dirichlet Allocation algorithm allows obtaining relevant topics such as violence applied through the use of weapons, family relationships within these organizations and the cultivation . of Cannabis. Furthermore, it is possible to detect non-trend concepts related to synthetic drugs where in recent times the use of 2CB, ketamine and fentanyl, which is currently going through a crisis in Chile, stand out.
Criminal organizations dedicated to drug trafficking are criminal groups that constantly look for the opportunity to commit their crimes and it is common knowledge that these events occur day after day and are reported in both popular and rural newspapers. The amount of police news is so large that evaluating the behavior of these organizations over the years can be very complex. This paper proposes a thematic analysis of a large number of police news articles since 2014 using unsupervised machine learning algorithms. The k-means clustering algorithm facilitates the observation of those groups of crimes corresponding to organized crime and the Latent Dirichlet Allocation algorithm allows obtaining relevant topics such as violence applied through the use of weapons, family relationships within these organizations and the cultivation . of Cannabis. Furthermore, it is possible to detect non-trend concepts related to synthetic drugs where in recent times the use of 2CB, ketamine and fentanyl, which is currently going through a crisis in Chile, stand out.
Description
Keywords
Algoritmos de aprendizaje, Organizaciones criminales, Algoritmos computacionales
