Logo
ScienceStack
Table of Contents
Fetching ...
Sign In

BengaliSent140: A Large-Scale Bengali Binary Sentiment Dataset for Hate and Non-Hate Speech Classification

Akif Islam, Sujan Kumar Roy, Md. Ekramul Hamid

TL;DR

BengaliSent140 tackles the scarcity of large scale Bengali annotated data for hate and non hate speech classification by consolidating seven heterogeneous datasets into a unified binary benchmark. It harmonizes diverse annotations into Hate and Not Hate labels and preserves multiple textual representations to study preprocessing effects, enabling robust deep learning benchmarking. Baseline experiments show transformer models like BERT achieving the highest accuracy (0.91), with other neural and classical methods providing competitive results, underscoring the dataset's utility for cross paradigm evaluation. Publicly released on Kaggle, BengaliSent140 offers broad linguistic coverage and a strong foundation for developing and evaluating Bengali hate speech and sentiment analysis systems in real-world, multi-domain contexts.

Abstract

Sentiment analysis for the Bengali language has attracted increasing research interest in recent years. However, progress remains constrained by the scarcity of large-scale and diverse annotated datasets. Although several Bengali sentiment and hate speech datasets are publicly available, most are limited in size or confined to a single domain, such as social media comments. Consequently, these resources are often insufficient for training modern deep learning based models, which require large volumes of heterogeneous data to learn robust and generalizable representations. In this work, we introduce BengaliSent140, a large-scale Bengali binary sentiment dataset constructed by consolidating seven existing Bengali text datasets into a unified corpus. To ensure consistency across sources, heterogeneous annotation schemes are systematically harmonized into a binary sentiment formulation with two classes: Not Hate (0) and Hate (1). The resulting dataset comprises 139,792 unique text samples, including 68,548 hate and 71,244 not-hate instances, yielding a relatively balanced class distribution. By integrating data from multiple sources and domains, BengaliSent140 offers broader linguistic and contextual coverage than existing Bengali sentiment datasets and provides a strong foundation for training and benchmarking deep learning models. Baseline experimental results are also reported to demonstrate the practical usability of the dataset. The dataset is publicly available at https://www.kaggle.com/datasets/akifislam/bengalisent140/

BengaliSent140: A Large-Scale Bengali Binary Sentiment Dataset for Hate and Non-Hate Speech Classification

TL;DR

BengaliSent140 tackles the scarcity of large scale Bengali annotated data for hate and non hate speech classification by consolidating seven heterogeneous datasets into a unified binary benchmark. It harmonizes diverse annotations into Hate and Not Hate labels and preserves multiple textual representations to study preprocessing effects, enabling robust deep learning benchmarking. Baseline experiments show transformer models like BERT achieving the highest accuracy (0.91), with other neural and classical methods providing competitive results, underscoring the dataset's utility for cross paradigm evaluation. Publicly released on Kaggle, BengaliSent140 offers broad linguistic coverage and a strong foundation for developing and evaluating Bengali hate speech and sentiment analysis systems in real-world, multi-domain contexts.

Abstract

Sentiment analysis for the Bengali language has attracted increasing research interest in recent years. However, progress remains constrained by the scarcity of large-scale and diverse annotated datasets. Although several Bengali sentiment and hate speech datasets are publicly available, most are limited in size or confined to a single domain, such as social media comments. Consequently, these resources are often insufficient for training modern deep learning based models, which require large volumes of heterogeneous data to learn robust and generalizable representations. In this work, we introduce BengaliSent140, a large-scale Bengali binary sentiment dataset constructed by consolidating seven existing Bengali text datasets into a unified corpus. To ensure consistency across sources, heterogeneous annotation schemes are systematically harmonized into a binary sentiment formulation with two classes: Not Hate (0) and Hate (1). The resulting dataset comprises 139,792 unique text samples, including 68,548 hate and 71,244 not-hate instances, yielding a relatively balanced class distribution. By integrating data from multiple sources and domains, BengaliSent140 offers broader linguistic and contextual coverage than existing Bengali sentiment datasets and provides a strong foundation for training and benchmarking deep learning models. Baseline experimental results are also reported to demonstrate the practical usability of the dataset. The dataset is publicly available at https://www.kaggle.com/datasets/akifislam/bengalisent140/
Paper Structure (13 sections, 3 figures, 4 tables)

This paper contains 13 sections, 3 figures, 4 tables.

Table of Contents

  1. Introduction
  2. Related Work
  3. Dataset Construction
  4. Source Datasets
  5. Data Preprocessing
  6. Experimental Setup
  7. Models
  8. Embedding Strategies
  9. Evaluation Metrics
  10. Results and Discussion
  11. Model Performance on BengaliSent140
  12. Challenges and Limitations
  13. Conclusion and Future Work

Figures (3)

  • Figure 1: Source-wise contribution of text samples from the seven datasets merged to construct the BengaliSent140 corpus, illustrating the relative size and diversity of each data source.
  • Figure 2: Binary class distribution of the BengaliSent140 dataset, illustrating the number of hate and not-hate text samples.
  • Figure 3: Illustration of semantic ambiguity in Bengali, where identical words convey different meanings depending on context, posing challenges for sentiment and hate speech classification.