abdelhaqueidali/Amazigh-Dictionary-Dataset

Dataset Card for Amazigh Scraped Dictionaries

This dataset is a collection of various Amazigh dictionaries scraped from their respective mobile applications. It is intended strictly for non-commercial research, learning, and the development of tools supporting the low-resource Amazigh language online.

Important Disclaimer: The compiler of this dataset does not claim ownership or copyright over this data. All rights remain with the original creators and institutions. If you are a rights holder and wish for your data to be removed, please contact the dataset compiler, and it will be taken down immediately.

Dataset Details

Dataset Description

This dataset aggregates multiple Amazigh lexical resources into a central repository to aid NLP training, research, and language support tools.

• Curated by: [Abdelhaque Id Ali]
• Language(s) (NLP): Amazigh (Standard), Arabic, French, English.
• License: Unknown / Other (All rights reserved to original authors).

Dataset Sources

• Repository: [https://huggingface.co/datasets/abdelhaqueidali/Amazigh-Dictionary-Dataset/]

Uses

Direct Use

This dataset is intended for anything requiring Amazigh word-to-word translations, including:

• Machine learning training for Amazigh Natural Language Processing (NLP).
• Academic and linguistic research.
• Learning and educational purposes.
• Building applications that support the Amazigh language.

Usage Warning: Each field of application requires careful consideration of copyright. Developers building public-facing applications are strongly advised to seek direct permission from the original data owners to ensure legal compliance.

Out-of-Scope Use

• Commercial use or monetization of the dictionary data without explicit owner consent.
• Claiming ownership of the dataset contents.

Chain of Rights and Licensing Caveat

Users must be aware of the "chain of rights" present in this dataset. Many of the third-party dictionaries included here (such as Tafsut and the msmunawal apps) are built upon the foundational dictionary provided by IRCAM.

Obtaining usage permission from the secondary developers (e.g., Fatima Ezzahra El Moussaffir or the Tafsut developers) may not be legally sufficient. Unless those developers secured the right to sub-license IRCAM's data, users of this dataset may still require explicit, direct permission from IRCAM for their specific use cases.

Dataset Structure

The dataset is divided into two main folders:

• sqlite-db: Contains the original scraped SQLite database files. This is the recommended source as it preserves the original structure.
• json: Contains a quick conversion of the databases into JSON format. Note: The JSON conversion was done without structural changes and may contain formatting issues. Users are advised to rely on the sqlite-db files and process them carefully to JSON if errors occur.

File List:

• dglai6.db, dglai7.db, dglai13.db, dglai14.db, dglai16.db, dglai17.db, dglai20.db
• msmun_ar.db, msmun_fr.db, new_msmunawal_list1.db, new_msmunawal_list2.db
• tafsut_en.db
• tawalt_ar.db, tawalt_fr.db

Dataset Creation

Curation Rationale

This data was collected to overcome the scarcity of accessible lexical data for Amazigh and to preserve access to tools that frequently go offline.

Source Data

Data Collection and Processing

The data was scraped "as-is" directly from the source applications. Known Limitations: The current IRCAM dglai datasets lack the annexation state of nouns (État d'annexion), which is available on their live website. The files reflect older app versions.

Who are the source data producers?

The data originates from the following creators. Please support their original platforms:

• IRCAM (Institut Royal de la Culture Amazighe)

  • Files: dglai versions (6, 7, 13, 14, 16, 17, 20)
  • Website: [http://www.ircam.ma]
  • LinkedIn: [Insert Link]

• Madghis U'Madi

  • Files: tawalt_ar.db, tawalt_fr.db
  • Email: [Insert Link]
  • LinkedIn: [Insert Link]
  • Social Media: [Insert Link]

• Fatima Ezzahra El Moussaffir

  • Files: msmun_ar.db, msmun_fr.db, new_msmunawal_list1.db, new_msmunawal_list2.db
  • Email: [Insert Link]
  • LinkedIn: [Insert Link]
  • Social Media: [Insert Link]

• Tafsut Developer

  • Files: tafsut_en.db
  • Note: Apps currently unavailable on the Play Store; no contact info available.

Bias, Risks, and Limitations

Users should be aware that dictionaries may contain inconsistencies or reflect specific regional dialects depending on the source author's background.

Dataset Card Contact

Takedown Requests & Contact: If you are a copyright holder and wish for your data to be removed from this repository, please contact me at: [abdelhaque379@gmail.com]. The files will be deleted without delay.