Publicação
Large Language Models Powered Aspect-Based Sentiment Analysis: A Revolution in the Obtention of Customer Insights
| Resumo: | In the age of social networks, user-generated content is crucial for businesses, directly influencing their decisions. Traditional sentiment analysis methods struggle with the vast amount of information and the implicit aspects of sentiments expressed. This study explores the potential of Aspect-Based Sentiment Analysis (ABSA) powered by Large Language Models (LLMs) to improve sentiment analysis. By employing GPT-4 through ChatGPT, we analyze a dataset of all-inclusive hotel reviews to benchmark three approaches: a fuzzy logic-based method, a human (manual) analysis, and a ChatGPT-based analysis. Our findings reveal a high similarity between ChatGPT and Human analysis, with ChatGPT demonstrating a high ability to interpret nuanced language and handle subjectivity. This study not only highlights LLMs' potential to automate and enhance customer sentiment analysis, providing businesses with deeper insights and improving their responsiveness to customer feedback but also contributes to academia by offering new insights into the application of LLMs in ABSA, paving the way for further research and development in this field. |
|---|---|
| Autores principais: | Água, Mariana Rodrigues Borda de |
| Assunto: | Sentiment Analysis Aspect-Based Sentiment Analysis Large Language Models ChatGPT GPT-4 SDG 9 - Industry, innovation and infrastructure |
| Ano: | 2024 |
| País: | Portugal |
| Tipo de documento: | dissertação de mestrado |
| Tipo de acesso: | acesso embargado |
| Instituição associada: | Universidade Nova de Lisboa |
| Idioma: | inglês |
| Origem: | Repositório Institucional da UNL |
| Resumo: | In the age of social networks, user-generated content is crucial for businesses, directly influencing their decisions. Traditional sentiment analysis methods struggle with the vast amount of information and the implicit aspects of sentiments expressed. This study explores the potential of Aspect-Based Sentiment Analysis (ABSA) powered by Large Language Models (LLMs) to improve sentiment analysis. By employing GPT-4 through ChatGPT, we analyze a dataset of all-inclusive hotel reviews to benchmark three approaches: a fuzzy logic-based method, a human (manual) analysis, and a ChatGPT-based analysis. Our findings reveal a high similarity between ChatGPT and Human analysis, with ChatGPT demonstrating a high ability to interpret nuanced language and handle subjectivity. This study not only highlights LLMs' potential to automate and enhance customer sentiment analysis, providing businesses with deeper insights and improving their responsiveness to customer feedback but also contributes to academia by offering new insights into the application of LLMs in ABSA, paving the way for further research and development in this field. |
|---|