---
license: mit
task_categories:
- text-generation
language:
- bn
tags:
- bengali
- poetry
- literature
- web-scraped
- non-ai
pretty_name: Bangla Kobita Poetry Archive
size_categories:
- 100K<n<1M
---

# Bangla Kobita Poetry Archive

## Overview
This repository contains a curated text dataset of Bengali poetry scraped from the web, primarily targeting comprehensive poetry platforms like [bangla-kobita.com](https://www.bangla-kobita.com/). The primary goal of this archive is to preserve a rich collection of purely human-written Bengali poems (Bangla Kobita), creating a distinct record of human artistic expression, emotion, and linguistic rhythm separate from AI-generated text.

## Purpose and Usage
This dataset is published publicly and strictly for **educational, research, linguistic analysis, and archival purposes**. It is a highly specialized resource for Natural Language Processing (NLP) researchers, data scientists, and linguists looking to:
* Pre-train or fine-tune Bengali Large Language Models (LLMs) on creative, metaphorical, and emotionally complex text.
* Study poetic structures, rhyme schemes, and metrical patterns in modern and classic Bengali literature.
* Perform sentiment analysis and emotion detection on deeply expressive artistic writing.

## Dataset Details
* **Source:** bangla-kobita.com (and/or related Bengali poetry archives)
* **Collection Method:** Web scraping
* **Content Type:** Text (Bengali poems, including titles, poet names, and stanza-formatted text written by humans without the use of AI).
* **Repository:** `sayurio/bangla-kobita-scrape-bangla-literature`

## Copyright and Fair Use Disclaimer
This archive is created under the principles of **Fair Use** (under Section 107 of the Copyright Act) for purposes such as criticism, comment, teaching, scholarship, and research.

* **No Ownership Claimed:** The creator of this repository does not claim any ownership, authorship, or copyright over the original poems. All rights, title, and interest in the original text remain entirely with their respective poets, authors, and the publishing platforms.
* **Non-Commercial:** This dataset is provided completely free of charge and is strictly not intended for commercial gain, monetization, or profit.
* **Transformative Use:** The data has been aggregated, extracted from its original web formatting, and compiled specifically for computational analysis, archiving, and educational study. This represents a transformative use of the original publicly available artistic material.

**Takedown Requests:** If you are a copyright holder or a poet whose work is included in this dataset and wish for it to be removed from this archive, please open an issue or contact the repository owner directly. Please submit a removal request specifying the exact poem titles, URLs, or text snippets you wish to have taken down so they can be accurately located within the dataset and removed.

## How to Use
You can load this dataset directly into your Python environment using the Hugging Face `datasets` library:

```python
from datasets import load_dataset

# Load the dataset
dataset = load_dataset("sayurio/bangla-kobita-scrape-bangla-literature")

# View the structure of the first poem entry
print(dataset['train'][0])
```