Fit Talks: Forecasting Fitness Awareness in Saudi Arabia Using Fine-Tuned Transformers

• DOI: 10.3390/bdcc9020020
• Publication Date: 2025-01-23T09:54:31Z
• Authors: Nora Alturayeif; Deemah Alqahtani; Sumayh S. Aljameel; Najla Almajed; Lama Alshehri; Nourah Aldhuwaihi; Madawi Alhadyan; Nouf Aldakheel
• Keywords: Technology; transformer-based models; sentiment analysis; T; deep learning; natural language processing

Understanding public sentiment on health and fitness is essential for addressing regional challenges in Saudi Arabia. This research employs analysis to assess awareness by analyzing content from the X platform (formerly Twitter), using a dataset called Aware, which includes 3593 posts related awareness. Preprocessing steps such as normalization, stop-word removal, tokenization ensured high-quality data. The findings revealed that positive sentiments about were more prevalent than negative ones, with across all categories being most common western region. However, eastern region exhibited highest percentage of sentiment, indicating strong interest health. For classification, we fine-tuned two transformer architectures—BERT GPT—utilizing three BERT-based models (AraBERT, MARBERT, CAMeLBERT) GPT-3.5. These provide valuable insights into Arabian attitudes toward health, offering actionable information campaigns initiatives.