Análisis de sentimientos en reseñas hoteleras de Buenaventura mediante minería de texto

Fecha
2026
Autores
Título de la revista
ISSN de la revista
Título del volumen
Editor
Universidad de Manizales
Referencia bibliográfica
Resumen
El análisis de sentimientos aplicado al turismo se ha consolidado como una herramienta fundamental para comprender las percepciones de los visitantes y orientar la toma de decisiones en destinos emergentes. En Buenaventura, ciudad con alto potencial turístico pero marcada por desafíos estructurales como la seguridad y la informalidad empresarial, no existían estudios que sistematizaran de manera rigurosa las opiniones disponibles en plataformas digitales.
Este proyecto examinó 3.727 reseñas publicadas en Google entre 2020 y 2025 sobre los cinco hoteles más comentados de la ciudad, mediante técnicas de minería de texto, análisis automatizado de sentimientos y visualización de datos. El proceso siguió el modelo CRISP-DM, articulando un scraping automatizado con Apify, un flujo de procesamiento en Azure basado en la arquitectura Bronze – Silver – Gold y un enriquecimiento semántico generado a través de Azure Cognitive Services – Text Analytics.
El estudio clasificó la polaridad emocional de las reseñas, identificó aspectos temáticos y reconoció tendencias narrativas asociadas a la experiencia turística. La validación manual de una muestra del 3% arrojó una coherencia interpretativa del 90,2%. Los resultados fueron integrados en dashboards interactivos en Power BI, conformando un sistema visual capaz de revelar patrones de satisfacción, alertas temáticas y brechas en la respuesta hotelera, con valor estratégico para turistas, empresarios y entidades locales.
Sentiment analysis applied to tourism has become a fundamental tool for understanding visitors’ perceptions and guiding decision-making in emerging destinations. In Buenaventura, a city with significant tourism potential but affected by structural challenges such as security conditions and business informality, no studies had rigorously systematized the opinions available on digital platforms. This project examined 3,727 reviews posted on Google between 2020 and 2025 concerning the five most frequently reviewed hotels in the city, using text mining techniques, automated sentiment analysis, and data visualization. The process followed the CRISP-DM model, integrating automated scraping through Apify, a processing workflow in Azure based on the Bronze–Silver–Gold architecture, and semantic enrichment generated through Azure Cognitive Services – Text Analytics. The study enabled the classification of emotions, the identification of thematic aspects, and the detection of narrative trends associated with the tourism experience. Manual validation of a 3% sample yielded an interpretative coherence of 90.2%. The findings were incorporated into interactive dashboards in Power BI, creating a visual system capable of revealing satisfaction patterns, thematic alerts, and gaps in hotel responses, offering strategic value for tourists, business owners, and local institutions.
Sentiment analysis applied to tourism has become a fundamental tool for understanding visitors’ perceptions and guiding decision-making in emerging destinations. In Buenaventura, a city with significant tourism potential but affected by structural challenges such as security conditions and business informality, no studies had rigorously systematized the opinions available on digital platforms. This project examined 3,727 reviews posted on Google between 2020 and 2025 concerning the five most frequently reviewed hotels in the city, using text mining techniques, automated sentiment analysis, and data visualization. The process followed the CRISP-DM model, integrating automated scraping through Apify, a processing workflow in Azure based on the Bronze–Silver–Gold architecture, and semantic enrichment generated through Azure Cognitive Services – Text Analytics. The study enabled the classification of emotions, the identification of thematic aspects, and the detection of narrative trends associated with the tourism experience. Manual validation of a 3% sample yielded an interpretative coherence of 90.2%. The findings were incorporated into interactive dashboards in Power BI, creating a visual system capable of revealing satisfaction patterns, thematic alerts, and gaps in hotel responses, offering strategic value for tourists, business owners, and local institutions.
Descripción
Palabras clave
Análitica de datos, Minería de Texto, Desarrollo informático, Turismo - Análisis Estadístico, Percepción Turística, Turismo - Buenaventura (Valle del Cauca-Colombia)