Obhibekti: A Multisource Benchmark Corpus for Bengali Emotion Classification

Published: 8 December 2025| Version 1 | DOI: 10.17632/ft69jmsnbc.1
Contributors:
Rabeya Mumu, Tashfia Hossain, Oindrila Saha, Morshed Alam, Shabnam Hasan

Description

The Obhibekti dataset consists of 3,928 Bengali texts collected from reliable and diverse sources such as online forums, news websites, and social media platforms like Facebook. The dataset was developed to represent six core human emotions: happiness, sadness, anger, fear, disgust, and surprise. The distribution of the six emotion categories is as follows: Happy 650 Sad: 650 Surprise: 650 Disgust: 650 Fear: 657 Anger: 674

Files

Institutions

  • Ahsanullah University of Science and Technology

Categories

Natural Language Processing, Machine Learning, Bangladesh, Deep Learning

Licence