Datasets:

mlibre
/

palestine

---
license: unknown
language:
- en
tags:
- Palestine
- dataset
pretty_name: Palestine
size_categories:
- 10M<n<100M
configs:
  - config_name: default
    data_files:
      - split: train
        path: combined/combined_with_metadata.jsonl
---
# Palestine Dataset 🇵🇸

A curated dataset focused on authentic Palestinian history, narratives and reporting.

## Data Sources 📊

* **decolonizepalestine.com** - Educational content and historical documentation
* **electronicintifada.net** - Hundreds of articles - news, analysis, and more
* **palianswers.com** - A crowdsourced database of short responses to Zionist claims
* **english.khamenei.ir** - Articles related to Palestine
* **mondoweiss.net** - Hundreds of articles - news, analysis, and more
* **stand-with-palestine.org** - Interviews and articles
* **bdsmovement.net** - Boycott, Divestment, and Sanctions (BDS) movement

## Purpose

This dataset aims to preserve and provide access to information about Palestine by collecting content from selected sources that represent Palestinian perspectives and history.

## Note on Sources

While this dataset strives to provide quality information from carefully selected sources, users are encouraged to:

* Cross-reference information
* Form their own informed conclusions
* Understand that historical events can have multiple perspectives

## Dataset Structure 📁

The repository is organized as follows:

* `combined` directory:
  * Combined data from all sources in multiple formats:
    * `combined.csv` and `combined.jsonl` - Basic combined data
    * `combined_with_metadata.csv` and `combined_with_metadata.jsonl` - Combined data with additional metadata
  * `texts` directory - Individual numbered text files
  * `texts_with_metadata` directory - Text files with metadata information

* `decolonizepalestine`, `electronicintifada`, `mondoweiss`, `palianswers`, `stand-with-palestine`, `khamenei-ir-free-palestine-tag`, `khamenei-ir-palestine-special-page`, `bdsmovement` are the directories for each source:
  * Includes raw `texts`, `CSV` and `JSONL` formats
  * Original `website content` preserved

Direct download link to `combined_with_metadata.jsonl` file: [combined_with_metadata.jsonl](https://huggingface.co/datasets/mlibre/palestine/resolve/main/combined/combined_with_metadata.jsonl)

## Contributing 🤝

### 1. Add New Sources 📚

* Submit trustworthy sources about Palestine through pull requests
* Sources can include news outlets, academic papers, historical archives
* Ensure sources provide authentic Palestinian perspectives
* Include source validation and credibility information

### 2. Model Training 🤖

* Use this dataset to fine-tune Language Models
* Create specialized models focused on Palestinian history and culture
* Help improve AI understanding of Palestinian narratives
* Share your training results and model performance metrics

### 3. Custom AI Applications 💡

* Develop custom GPTs using this dataset
* Create educational tools and chatbots
* Build applications that help spread awareness
* Share your applications and use cases with the community

Feel free to start contributing in any of these areas. Together we can help preserve and share Palestinian knowledge and history. 🌟

## Contact 📧

* <m.gh@linuxmail.org>
* <mlibrego@gmail.com>
* <https://github.com/mlibre>

## How to regenerate the dataset

The dataset is generated using the [Clean-Web-Scraper](https://github.com/mlibre/Clean-Web-Scraper) library.

```bash
git clone https://github.com/mlibre/Clean-Web-Scraper
cd Clean-Web-Scraper
npm install --ignore-scripts
node example-usage.js
```

## License ⚖️

* unlicense
* Content from source websites remains under their respective original licenses.