Conference Paper2024Cited by 1

Using Transformers for Emotion Recognition in Bangla Text: A Comparative Study of MultiBERT and BanglaBERT with Data Augmentation

N. Halder, T. P. Proma, J. H. Setu, A. Noor, A. Islam, M. A. Amin

23rd IEEE International Conference on Machine Learning and Applications (ICMLA)

IEEE, pp. 1657-1662

DOI: 10.1109/icmla61862.2024.00256URL

Abstract

Emotion recognition is an exacting task due to the complexity and diversity of human emotions, as well as their tendency to be open to multiple interpretations. Despite these challenges, emotion recognition from textual data has achieved promising results in recent years for the English language. Numerous studies have been conducted on this topic, leading to significant advancements in various fields, including business, politics, education, research, and healthcare. However, accurate emotion recognition from Bengali (Bangla) text remains difficult due to limited resources available for the Bangla language. Additionally, the available datasets for Bengali text are often poorly annotated or imbalanced, which significantly hampers the performance of text classifiers and causes overfitting. Data augmentation can help address this issue. By addressing class imbalance through data augmentation via back-translation, we analyzed and evaluated the impact of augmentation on both real and synthetic datasets. We used transformer-based models, Multi-BERT and BanglaBERT. Data augmentation considerably boosts the accuracy by 8% and 6% for MultiBERT and BanglaBERT respectively, along with a 13% increase in F1-score for both the models. An ablation study also evaluated the effects of hyperparameters on our models.

Keywords

  • BanglaBERT
  • back-translation
  • class imbalance
  • low-resource languages
  • data augmentation
  • Bengali text
  • hyperparameter effects
  • transformer models
  • synthetic dataset

References

  1. 1.Jakubův, Jan, Chvalovský, Karel, Goertzel, Zarathustra, Kaliszyk, Cezary, Olšák, Mirek, Piotrowski, Bartosz, Schulz, Stephan, Suda, Martin, Urban, Josef. (2023). Exploiting Generative AI to Scale up Intelligent Tutoring Systems. DROPS (Schloss Dagstuhl – Leibniz Center for Informatics)[10.4230/lipics.itp.2023.19]
  2. 2.Connor Shorten, Taghi M. Khoshgoftaar. (2019). A survey on Image Data Augmentation for Deep Learning. Journal Of Big Data, 6(1)[10.1186/s40537-019-0197-0]
  3. 3.Xing Fang, Justin Zhan. (2015). Sentiment analysis using product review data. Journal Of Big Data, 2(1)[10.1186/s40537-015-0015-2]
  4. 4.Željko Ð. Vujovic. (2021). Classification Model Evaluation Metrics. International Journal of Advanced Computer Science and Applications, 12(6)[10.14569/ijacsa.2021.0120670]
  5. 5.M. Maithri, U. Raghavendra, Anjan Gudigar, Jyothi Samanth, Prabal Datta Barua, Murugappan Murugappan, Yashas Chakole, U. Rajendra Acharya. (2022). Automated emotion recognition: Current trends and future perspectives. Computer Methods and Programs in Biomedicine, 215, 106646[10.1016/j.cmpb.2022.106646]
  6. 6.Amane Sugiyama, Naoki Yoshinaga. (2019). Data augmentation using back-translation for context-aware neural machine translation. , 35–44[10.18653/v1/d19-6504]
  7. 7.Sara Azmin, Kingshuk Dhar. (2019). Emotion Detection from Bangla Text Corpus Using Naïve Bayes Classifier. 2019 4th International Conference on Electrical Information and Communication Technology (EICT), 1–5[10.1109/eict48899.2019.9068797]
  8. 8.Mohammad Tareq, Md. Fokhrul Islam, Swakshar Deb, Sejuti Rahman, Abdullah Al Mahmud. (2023). Data-Augmentation for Bangla-English Code-Mixed Sentiment Analysis: Enhancing Cross Linguistic Contextual Understanding. IEEE Access, 11, 51657–51671[10.1109/access.2023.3277787]
  9. 9.María Lucia Barron-Estrada, Ramón Zatarain-Cabada, Raúl Oramas-Bustillos. (2019). Emotion Recognition for Education using Sentiment Analysis. Research in Computing Science, 148(5), 71–80[10.13053/rcs-148-5-8]
  10. 10.Mohsinul Kabir, Obayed Bin Mahfuz, Syed Rifat Raiyan, Hasan Mahmud, Md Kamrul Hasan. (2023). BanglaBook: A Large-scale Bangla Dataset for Sentiment Analysis from Book Reviews. , 1237–1247[10.18653/v1/2023.findings-acl.80]
  11. 11.Asmita De, Sushruta Mishra. (2022). Augmented Intelligence in Mental Health Care: Sentiment Analysis and Emotion Detection with Health Care Perspective. Studies in computational intelligence, 205–235[10.1007/978-981-19-1076-0_12]
  12. 12.Hasan Abid Ruposh, Mohammed Moshiul Hoque. (2019). A Computational Approach of Recognizing Emotion from Bengali Texts. , 570–574[10.1109/icaee48663.2019.8975417]
  13. 13.Simon Islam, Animesh Chandra Roy, Mohammad Shamsul Arefin, Sonia Afroz. (2021). Multi-label Emotion Classification of Tweets Using Machine Learning. Lecture notes on data engineering and communications technologies, 705–722[10.1007/978-981-16-6636-0_53]
  14. 14.Anna Koufakou, Diego Grisales, Ragy Costa de jesus, Oscar Fox. (2023). Data Augmentation for Emotion Detection in Small Imbalanced Text Data. , 1508–1513[10.1109/icmla58977.2023.00227]
  15. 15.Farhan Sadaf, Abdul Muntakim, K. M. Azharul Hasan. (2024). Building an Affective Database for Emotion Detection from Natural Bangla Text. Lecture notes in networks and systems, 621–635[10.1007/978-981-99-8937-9_42]
  16. 16.Rajesh Kumar Das, Samrina Sarkar Sammi, Khadijatul Kobra, Moshfiqur Rahman Ajmain, Sharun Akter khushbu, Sheak Rashed Haider Noori. (2023). Analysis of Bangla Transformation of Sentences Using Machine Learning. Lecture notes in networks and systems, 36–52[10.1007/978-3-031-30396-8_4]

Cited By

  1. 1.Kahakashan Ashraf, Atiya Masuda Siddika, Mohammad Shamsul Arefin. (2026). Analyzing Arrogant Comments in the Bengali Language Through Explainable AI and NLP. , 1–6[10.1109/icecte69292.2026.11429355]
← All publications