Sistema de detección de transacciones atípicas apoyado por inteligencia artificial
Loading...
Date
2025-12-17
Authors
Journal Title
Journal ISSN
Volume Title
Publisher
Centro Universitario Tecnológico CEUTEC
Abstract
El presente proyecto de graduación se enfocó específicamente en el reciente incremento de transacciones de fraude con tarjetas de crédito en instituciones financieras medianas y pequeñas de Honduras. Este fenómeno ha afectado la salud laboral y operativa de los departamentos de riesgo transaccional, al verse sobrecargados por el monitoreo manual de alertas y la falta de herramientas tecnológicas adecuadas para identificar patrones sofisticados de fraude. La metodología utilizada fue de enfoque mixto, combinando entrevistas cualitativas y encuestas aplicadas a especialistas de riesgo transaccional y oficiales de alerta de dos instituciones financieras representativas. Los resultados obtenidos revelan que, si bien el 100% de los encuestados utilizan herramientas tecnológicas, estas carecen de capacidades avanzadas de detección, lo que limita su capacidad de adaptación ante fraudes emergentes. Ante este escenario, el proyecto propone el diseño de una herramienta antifraude parametrizable que utilice modelos de inteligencia artificial para detectar transacciones sospechosas en tiempo real. Esta solución está diseñada para ser operada por instituciones con recursos limitados, sin necesidad de contar con equipos especializados en IA. Mejorando de esta manera la productividad del área de riesgo transaccional, la cual se enfoca en mitigar el fraude con tarjetas de crédito en estas instituciones bancarias medianas o pequeñas del país. Palabras clave: Productividad, salud laboral, herramienta antifraude.
This graduation project specifically focused on the recent increase in credit card fraud transactions in medium and small financial institutions in Honduras. This phenomenon has negatively impacted the operational and occupational well-being of transactional risk departments, which are overwhelmed by the manual monitoring of alerts and the lack of adequate technological tools to identify sophisticated fraud patterns. The methodology used followed a mixed-methods approach, combining qualitative interviews and surveys applied to transactional risk specialists and alert officers from two representative financial institutions. The results revealed that although 100% of the respondents use technological tools, these lack advanced detection capabilities, limiting their ability to adapt to emerging fraud schemes. In response to this scenario, the project proposes the design of a customizable anti-fraud tool that uses artificial intelligence models to detect suspicious transactions in real time. This solution is designed to be operated by institutions with limited resources, without the need for specialized AI teams. In doing so, it aims to enhance the productivity of transactional risk departments, which are focused on mitigating credit card fraud in these small and medium-sized financial institutions across the country.
This graduation project specifically focused on the recent increase in credit card fraud transactions in medium and small financial institutions in Honduras. This phenomenon has negatively impacted the operational and occupational well-being of transactional risk departments, which are overwhelmed by the manual monitoring of alerts and the lack of adequate technological tools to identify sophisticated fraud patterns. The methodology used followed a mixed-methods approach, combining qualitative interviews and surveys applied to transactional risk specialists and alert officers from two representative financial institutions. The results revealed that although 100% of the respondents use technological tools, these lack advanced detection capabilities, limiting their ability to adapt to emerging fraud schemes. In response to this scenario, the project proposes the design of a customizable anti-fraud tool that uses artificial intelligence models to detect suspicious transactions in real time. This solution is designed to be operated by institutions with limited resources, without the need for specialized AI teams. In doing so, it aims to enhance the productivity of transactional risk departments, which are focused on mitigating credit card fraud in these small and medium-sized financial institutions across the country.
Keywords
Productividad, Herramientas Antifraude, Salud Laboral