A Comparative Study of LSTM and Bi-LSTM Architectures with Attention for Bangla News Classification

DOI

10.2991/978-94-6239-664-7_31How to use a DOI?

Keywords

LSTM; Bi-LSTM; Bangla News Classification; Natural Language Processing; Attention Mechanism; Computational Linguistics

Abstract

News classification is an important task to perform in NLP, more so when dealing with low-resource languages such as Bangla. However, Bangla comes with its own set of challenges like different morphology, complex syntax, and a very acute shortage of large annotated corpora. In this work, Long Short-Term Memory (LSTM), Bidirectional LSTM (Bi-LSTM), LSTM with Attention, and Bi-LSTM with Attention were tested to classify Bangla news articles into four broad categories: sports, national, international, and entertainment. The balanced Kaggle dataset has 11,904 labeled samples. The dataset underwent preprocessing that included normalization, tokenization, padding, and removal of stopwords. Models were trained using the Adam optimizer for twenty epochs on categorical crossentropy loss. Results showed that Bi-LSTM + Attention set the highest validation accuracy of 96%, outperforming all other models. These results have paved the way that attention-based deep learning models can effectively classify Bangla news domains.

Copyright

© 2026 The Author(s)

Open Access

Open Access This chapter is licensed under the terms of the Creative Commons Attribution-NonCommercial 4.0 International License (http://creativecommons.org/licenses/by-nc/4.0/), which permits any noncommercial use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons license and indicate if changes were made.

Download article (PDF)

TY  - CONF
AU  - Md. Shazedur Rahman
AU  - Parvaj Kazi
AU  - Mir Mynul Ahasan Mim
AU  - Sadman Sadik Khan
AU  - Nitta Nando Roy
AU  - Md Anisur Rahman
PY  - 2026
DA  - 2026/06/08
TI  - A Comparative Study of LSTM and Bi-LSTM Architectures with Attention for Bangla News Classification
BT  - Proceedings of the International Conference on Intelligent Data Analysis and Applications (IDAA 2025)
PB  - Atlantis Press
SP  - 445
EP  - 458
SN  - 1951-6851
UR  - https://doi.org/10.2991/978-94-6239-664-7_31
DO  - 10.2991/978-94-6239-664-7_31
ID  - Rahman2026
ER  -