Explainable Bangla Linguistic Style Classification into Saint and Common Forms: A Deep Neural Network and Transformer-Based Approach with LIME-Based Interpretability
This study presents a deep neural network and transformer-based framework for classifying Bengali texts into saint and common forms, with SahajBERT achieving the best performance. The integration of LIME enhances interpretability, providing insights into stylistic cues aligned with linguistic expectations.
Oct 28, 2025