README.md

metadata

dataset_info:
  features:
    - name: title
      dtype: string
    - name: url
      dtype: string
    - name: content
      dtype: string
    - name: summary_content
      dtype: string
  splits:
    - name: train
      num_bytes: 298913247.43590635
      num_examples: 45995
    - name: test
      num_bytes: 74729936.56409365
      num_examples: 11499
  download_size: 202695419
  dataset_size: 373643184
configs:
  - config_name: default
    data_files:
      - split: train
        path: data/train-*
      - split: test
        path: data/test-*
task_categories:
  - sentence-similarity
  - text-classification
  - summarization
language:
  - id
tags:
  - islam
  - indonesia
  - news
  - article
size_categories:
  - 10K<n<100K

Indo-Islamic Article Dataset

The Indo-Islamic Article Dataset contains Islamic articles in Indonesian, collected from various web sources. The dataset supports tasks such as text classification, summarization, and sentence similarity. It was primarily used in the SEQURAN project, an Islamic knowledge-based search engine focusing on information retrieval and summarization.

Columns:

• title: The title of the article.
• url: The source URL where the article was published.
• content: Full content of the article.
• summary_content: Automatically generated summaries of the articles using bert-extractive-summarizer and the model ramadita/indo-islamic-sentence-bert-v2.

Creators:
Ray Ramadita and Muhammad Rizki Al-Fathir. Special thanks to Muhammad Rizki Al-Fathir for contributing over 10,000 articles.

Sources:
The articles in this dataset were collected from the following websites:

• Bimbingan Islam
• Bincang Syariah
• Bersama Dakwah
• Belajar Islam
• Dalam Islam
• Detik
• Eramuslim
• Hidayatullah
• iNews
• IslamQA
• Konsultasi Syariah
• Laduni
• Muhammadiyah
• NU Online
• Pesantren Al Irsyad
• Radio Mutiara Quran
• Suara Islam
• Suara Muhammadiyah
• Fimadani
• Harakatuna
• Peci Hitam
• PISS KTB
• Dakwah
• Islami
• Muslim.or.id
• Rumaysho
• Doa Harian Islami
• Fatwa Tarjih

Usage:
This dataset is highly useful for NLP tasks, particularly in the Islamic domain. The dataset includes 45,995 training examples and 11,499 testing examples. It can be used for text classification, summarization, and sentence similarity.

Citation:

If you use this dataset, please cite it as follows:

@dataset{ramadita_indo_islamic_article,
  author    = {Ray Ramadita and Muhammad Rizki Al-Fathir},
  title     = {Indo-Islamic Article Dataset},
  year      = {2024},
  url       = {https://huggingface.co/datasets/ramadita/indo-islamic-article},
}