Minería de opinión aplicada a comentarios de clientes en el sector turístico de Cuba
Loading...
Date
Authors
Journal Title
Journal ISSN
Volume Title
Publisher
Universidad de Sancti Spíritus "José Martí Pérez"
Abstract
El desarrollo digital y las redes sociales han transformado el turismo al permitir una conexión
rápida y democratizar la comunicación. A través de plataformas como Facebook, Instagram y
TripAdvisor, etc, los usuarios comparten opiniones que influyen en el sector turístico, lo que ha
llevado a la necesidad de herramientas que analicen y clasifiquen esta información. El en procesamiento
de estos datos se destaca el análisis de sentimiento o la minería de opinión como
disciplina que permite extraer y evaluar comentarios en entornos digitales, mejorando así la experiencia
del cliente y la reputación del sector turístico. En este contexto, se busca entender mejor
las opiniones de los viajeros para ajustar los servicios, y se plantea el desarrollo de un sistema que
utilice técnicas de minería de opinión y aprendizaje automático para clasificar los comentarios de
los clientes y extraer palabras clave, favoreciendo así una mejora continua en el sector turístico.
En el presente trabajo se traza como objetivo desarrollar un sistema que facilite la extracción
de palabras clave y la clasificación de la polaridad de las opiniones de los usuarios del sector
turístico de la región de Trinidad en la provincia de Sancti Spiritus en Cuba, utilizando técnicas
de minería de opinión y aprendizaje automático, así como modelos LLM para el Procesamiento
del Lenguaje Natural, extraídos de la plataforma Hugging Face. Este enfoque busca optimizar
el análisis de comentarios, mejorando la comprensión de las percepciones de los usuarios sobre
diversos temas. Como resultado se lograron optimizar varios modelos LLM altamente efectivos
para clasificar con precisión las opiniones de los usuarios y extraer palabras clave relevantes. Se
llevaron a cabo pruebas exhaustivas utilizando diferentes modelos, destacando su capacidad para
discernir entre comentarios positivos, negativos y neutros. Finalmente se implementa un sistema
basado en API para facilitar el acceso a estas herramientas analíticas a profesionales del sector
turístico y que puedan utilizarlas para tomar decisiones informadas.
The digital development and social networks have transformed tourism by enabling quick connections and democratizing communication. Through platforms like Facebook, Instagram, and TripAdvisor, users share opinions that influence the tourism sector, leading to the need for tools that analyze and classify this information. In processing these data, sentiment analysis or opinion mining stands out as a discipline that allows for the extraction and evaluation of comments in digital environments, thereby enhancing customer experience and the reputation of the tourism sector. In this context, the aim is to better understand travelers’ opinions in order to adjust services, and the development of a system is proposed that utilizes opinion mining techniques and machine learning to classify customer comments and extract keywords, thereby promoting continuous improvement in the tourism sector. The present work aims to develop a system that facilitates the extraction of keywords and the classification of the polarity of opinions from users in the tourism sector of the Trinidad region in the province of Sancti Spiritus in Cuba, using opinion mining techniques and machine learning, as well as LLM models for Natural Language Processing, sourced from the Hugging Face platform. This approach aims to optimize the analysis of comments, enhancing the understanding of users’ perceptions on various topics. As a result, several highly effective LLM models were optimized to accurately classify user opinions and extract relevant keywords. Extensive testing was conducted using different models, highlighting their ability to discern between positive, negative, and neutral comments. Finally, an API-based system was implemented to facilitate access to these analytical tools for professionals in the tourism sector, enabling them to make informed decisions.
The digital development and social networks have transformed tourism by enabling quick connections and democratizing communication. Through platforms like Facebook, Instagram, and TripAdvisor, users share opinions that influence the tourism sector, leading to the need for tools that analyze and classify this information. In processing these data, sentiment analysis or opinion mining stands out as a discipline that allows for the extraction and evaluation of comments in digital environments, thereby enhancing customer experience and the reputation of the tourism sector. In this context, the aim is to better understand travelers’ opinions in order to adjust services, and the development of a system is proposed that utilizes opinion mining techniques and machine learning to classify customer comments and extract keywords, thereby promoting continuous improvement in the tourism sector. The present work aims to develop a system that facilitates the extraction of keywords and the classification of the polarity of opinions from users in the tourism sector of the Trinidad region in the province of Sancti Spiritus in Cuba, using opinion mining techniques and machine learning, as well as LLM models for Natural Language Processing, sourced from the Hugging Face platform. This approach aims to optimize the analysis of comments, enhancing the understanding of users’ perceptions on various topics. As a result, several highly effective LLM models were optimized to accurately classify user opinions and extract relevant keywords. Extensive testing was conducted using different models, highlighting their ability to discern between positive, negative, and neutral comments. Finally, an API-based system was implemented to facilitate access to these analytical tools for professionals in the tourism sector, enabling them to make informed decisions.