A Review of State-of-the-Art Transformers for Bengali Text Summarization

This comprehensive review examines both Extractive and Abstractive approaches to Bengali Text Summarization (BTS). While deep learning has significantly advanced summarization for resource-rich languages like English, Bengali remains largely underexplored. The review covers BTS research from 2007 to 2023, analyzing trends, datasets, preprocessing methods, methodologies, evaluations, and challenges. It draws from 106 journal and conference papers to offer insights into emerging topics in Bengali Abstractive summarization. The study also augments the review with experiments using transformer models from Hugging Face and publicly available datasets to assess ROUGE score accuracy.