Fuzzy-BERT Synergy: A Multilayer Framework for Emotion Narrative in Hybrid Text Classification Models

This study introduces Fuzzy-BERT Synergy, a multilayer framework that integrates the capabilities of the IndoBERT model with a fuzzy logic system to analyze emotions and their intensity in Indonesian digital tales. The proposed methodology comprises three primary components: emotion classification using IndoBERT, emotion intensity evaluation using fuzzy inference, and quantitative assessment of narrative strength (Story Level). An assessment of 110 digital narratives indicated that this approach effectively identified intricate emotional subtleties, demonstrating a substantial link between fuzzy emotion intensity and the Story Level (r = 0.68). The findings also demonstrate a strong correlation between narrative strength and user engagement metrics, including likes and comments. This framework's advantage is its capacity to detect emotional gradients overlooked by traditional models and its responsiveness to alterations in story structure. These findings provide prospects for the advancement of story recommendation systems, automated narrative quality evaluation, and applications in gaming and digital story-based education. This study emphasizes the necessity for additional validation of multilingual data and various narrative genres to broaden the model's applicability.