BanglaTLit: A Benchmark Dataset for Back-Transliteration of Romanized Bangla
Md Fahim, Fariha Tanjim Shifat, Fabiha Haider, Deeparghya Dutta Barua, MD Sakib Ul Rahman Sourove, Md Farhan Ishmam, Md Farhad Alam Bhuiyan
Abstract
Low-resource languages like Bangla are severely limited by the lack of datasets. Romanized Bangla texts are ubiquitous on the internet, offering a rich source of data for Bangla NLP tasks and extending the available data sources. However, due to the informal nature of romanized text, they often lack the structure and consistency needed to provide insights. We address these challenges by proposing: (1) BanglaTLit, the large-scale Bangla transliteration dataset consisting of 42.7k samples, (2) BanglaTLit-PT, a pre-training corpus on romanized Bangla with 245.7k samples, (3) encoders further-pretrained on BanglaTLit-PT achieving state-of-the-art performance in several romanized Bangla classification tasks, and (4) multiple back-transliteration baseline methods, including a novel encoder-decoder architecture using further pre-trained encoders. Our results show the potential of automated Bangla back-transliteration in utilizing the untapped sources of romanized Bangla to enrich this language. The code and datasets are publicly available: https://github.com/farhanishmam/BanglaTLit.
BibTeX
@inproceedings{fahim-etal-2024-banglatlit,
title = "{B}angla{TL}it: A Benchmark Dataset for Back-Transliteration of {R}omanized {B}angla",
author = "Fahim, Md and
Shifat, Fariha Tanjim and
Haider, Fabiha and
Barua, Deeparghya Dutta and
Sourove, MD Sakib Ul Rahman and
Ishmam, Md Farhan and
Bhuiyan, Md Farhad Alam",
editor = "Al-Onaizan, Yaser and
Bansal, Mohit and
Chen, Yun-Nung",
booktitle = "Findings of the Association for Computational Linguistics: EMNLP 2024",
month = nov,
year = "2024",
address = "Miami, Florida, USA",
publisher = "Association for Computational Linguistics",
url = "https://aclanthology.org/2024.findings-emnlp.859/",
doi = "10.18653/v1/2024.findings-emnlp.859",
pages = "14656--14672"
}