axmeeabdhullo/axya-tech-websearch

Dhivehi Combined Dataset

Overview

This dataset combines three 30K Dhivehi language corpora from the Leipzig Corpora Collection into a single unified CSV file containing 90,000 sentences. The dataset provides a comprehensive resource for Dhivehi language processing, combining data from Wikipedia, news sources, and web crawls.

Dataset Composition

The dataset consists of three distinct sources:

• Wikipedia (2021): 30,000 sentences from Dhivehi Wikipedia
• News (2019): 30,000 sentences from Dhivehi news sources
• Web/Newscrawl (2015): 30,000 sentences from web crawl data

Total: 90,000 Dhivehi sentences

File Structure

• File: dhivehi_combined_dataset.csv
• Size: ~21MB
• Encoding: UTF-8
• Format: CSV with headers

CSV Columns

Column	Description
global_id	Sequential ID for the combined dataset (1-90000)
original_id	Original sentence ID from source dataset
sentence	The Dhivehi sentence text
source	Source dataset (Wikipedia, News, or Web)

Dataset Preview

You can explore the dataset using the Dataset Viewer above or the Data Studio feature on this page. Sample entries include:

• Sentences from various domains including religion, politics, history, medicine, and daily life
• Text in Thaana script (Dhivehi writing system)
• Diverse sentence lengths and structures

Download Instructions

Option 1: Direct Download

Download the CSV file directly from the Files tab of this dataset.

Option 2: Using Hugging Face Datasets Library

from datasets import load_dataset

# Load the dataset
dataset = load_dataset("axmeeabdhullo/axya-tech-websearch")

# Access the data
print(dataset['train'][0])

Option 3: Using pandas

import pandas as pd

# Load directly from Hugging Face
url = "https://huggingface.co/datasets/axmeeabdhullo/axya-tech-websearch/resolve/main/dhivehi_combined_dataset.csv"
df = pd.read_csv(url)

print(df.head())

Usage

This dataset can be used for:

• Dhivehi language modeling: Train language models for Dhivehi text generation
• Text analysis and NLP research: Analyze linguistic patterns in Dhivehi
• Machine translation training: Build translation systems for Dhivehi
• Linguistic corpus studies: Research Dhivehi language structure and usage
• Text classification: Train classifiers for Dhivehi text
• Summarization: Develop summarization models for Dhivehi content

Source Information

Leipzig Corpora Collection: https://wortschatz.uni-leipzig.de/en/download/Dhivehi

Original Data Sources:

• div_wikipedia_2021_30K
• div_news_2019_30K
• div_newscrawl_2015_30K

Licensing

This dataset is released under the Apache License 2.0.

You are free to:

• ✅ Use the dataset for commercial and non-commercial purposes
• ✅ Modify and distribute the dataset
• ✅ Use the dataset in your research and applications

Attribution: When using this dataset, please cite both this dataset and the Leipzig Corpora Collection (see Citation section below).

For full license details, see the Apache License 2.0.

Citation

If you use this dataset in your research or applications, please cite:

This Dataset

@dataset{dhivehi_combined_2024,
  title={Dhivehi Combined Dataset},
  author={axmeeabdhullo},
  year={2024},
  publisher={Hugging Face},
  url={https://huggingface.co/datasets/axmeeabdhullo/axya-tech-websearch}
}

Leipzig Corpora Collection

@inproceedings{goldhahn2012building,
  title={Building Large Monolingual Dictionaries at the Leipzig Corpora Collection: From 100 to 200 Languages},
  author={Goldhahn, Dirk and Eckart, Thomas and Quasthoff, Uwe},
  booktitle={Proceedings of the Eighth International Conference on Language Resources and Evaluation (LREC'12)},
  year={2012},
  pages={759--765}
}

Acknowledgments

This dataset was created by combining data from the Leipzig Corpora Collection, a project by the Natural Language Processing Group at Leipzig University. We gratefully acknowledge their work in collecting and providing access to corpora for low-resource languages like Dhivehi.

Special thanks to:

• The Leipzig Corpora Collection team for making Dhivehi language data available
• Contributors to Dhivehi Wikipedia
• Dhivehi news organizations that provided content
• The Hugging Face community for hosting and supporting dataset sharing

Contact & Support

For questions, issues, or suggestions regarding this dataset:

• Dataset Creator: @axmeeabdhullo
• Discussions: Use the Community tab to ask questions or share insights
• Issues: Report problems or request features through the Discussions tab

For issues related to the original Leipzig Corpora Collection data, please contact the Leipzig University team directly through their website.

Dataset Statistics

• Total Sentences: 90,000
• Language: Dhivehi (Divehi)
• Script: Thaana
• File Size: ~21MB
• Sources: 3 (Wikipedia, News, Web)
• Date Range: 2015-2021

---

Last Updated: September 2024