Título: LlaMA for Sentiment Analysis in Self-Care Applications: A Case Study with Life Energy
Autores: Inácio L. Souza Filho, Maryane de Castro Lima, Anderson I. M. Ripardo, Jessyca A. Bessa, Wedja J.G.Costa & Jermana Lopes de Moraes
Resumo: In recent years, self-care, and well-being have received growing attention, reflecting the importance in maintaining mental and emotional health amid the complexities of modern life. In this context, Life Energy emerges as an innovative tool designed to foster personal reflection and strengthen interpersonal connections. Structured around interactive tasks that explore physical, mental, and social dimensions, the instrument allows individuals to engage in a journey of self-discovery. The integration of emerging technologies, such as Artificial Intelligence, has further expanded the possibilities for mental health promotion through personalized interventions. This study presents the application of the LLaMA model for sentiment analysis within the Life Energy instrument, aiming to interpret user input, assess emotional states, and generate introspective questions. The proposed Life Energy Sentiment Model (LESM) was evaluated using both the Life Energy dataset and two public benchmark datasets: Amazon and IMDB reviews. The results for the Life Energy data were promising, especially regarding the generation of introspective questions that support user self-reflection. In the public datasets, the LESM achieved high performance, with accuracy, recall, and F1-scores all greater than 0.95, demonstrating its robustness and generalization capability across domains. These findings highlight the potential of combining large language models and structured self-care tools to create intelligent systems that promote emotional insight and personal development.
Palavras-chave: Self-awareness; Well-being; Sentiment Analysis; Large Language Models (LLMs); LLaMA; Artificial Intelligence; Human-Centered AI.
Páginas: 6
Código DOI: 10.21528/CBIC2025-1134393
Artigo em PDF: CBIC_2025_paper1134393.pdf
Arquivo BibTeX:
CBIC_2025_1134393.bib