Análisis de sentimientos en comentarios de TikTok sobre chatbots de IA generativa mediante Web Scraping y modelos Transformer

• Autores: Villena Cabrejos, Carlos Ricardo, Jauregui Romero, Eduardo Rafael
• Formato: artículo
• Fecha de Publicación: 2025
• Institución: Universidad Nacional Mayor de San Marcos
• Repositorio: Revistas - Universidad Nacional Mayor de San Marcos
• Lenguaje: español
• OAI Identifier: oai:revistasinvestigacion.unmsm.edu.pe:article/31314
• Enlace del recurso: https://revistasinvestigacion.unmsm.edu.pe/index.php/rpcsis/article/view/31314
• Nivel de acceso: acceso abierto
• Materia: Análisis de sentimientos; emociones; chatbots; TikTok; percepción del usuario; modelos transformadores; inteligencia artificial generativa; sentiment analysis; emotions; user perception; transformer models; generative artificial intelligence

In recent years, the rise of generative artificial intelligence chatbots on digital platforms has raised new questions about how users perceive and interact with these technologies. While technical progress has been made in chatbot development, a gap remains in understanding the emotions and attitudes they elicit—particularly in spontaneous digital environments such as TikTok. This study addresses this issue through an automated analysis of sentiments and feelings expressed in 76,645 TikTok comments about various generative AI chatbots (ChatGPT, Gemini, Claude, Perplexity, Copilot, Grok, and LLaMA), utilizing pre-trained Spanish transformer models. Quantitative results indicate a predominance of positive evaluations (44.89%) over neutral (30.93%) and negative (24.18%) ones. Emotions such as trust, joy, and optimism were the most frequent, although variations were detected across chatbot models. The findings suggest a generally favorable perception of generative AI chatbots on TikTok, while also highlighting significant differences in user opinions and emotional responses toward each system.