README.md

metadata

license: cc-by-4.0
task_categories:
  - token-classification
  - text-retrieval
  - text-generation
language:
  - wo
tags:
  - NER,
  - entity-linking
  - low-resource-nlp
  - wikidata
  - wolof
size_categories:
  - 1K<n<10K
pretty_name: Wolof Entity Linking

Dataset Summary

Entity Linking for low-resource African languages faces a major bottleneck: orthographic variation.

This gold standard dataset contains:

• 1,045 sentences (filtered from MasakhaNER 2.0 validation and test splits to exclude sentences without entity mentions or with only DATE mentions).
• 2,049 entity mentions categorized into LOC, ORG, and PER.
• 1,704 linkable mentions (In-KB) resolved to 565 unique Wikidata QIDs.
• 345 NIL mentions (Out-of-KB) representing entities absent from Wikidata.

Dataset Statistics

Entity Type	Linkable (In-KB)	NIL (Out-of-KB)	Total Mentions	Unique QIDs
Locations (LOC)	758	23	781	174
Organizations (ORG)	420	128	548	173
Persons (PER)	526	194	720	222
Total	1,704	345	2,049	565

Dataset Structure

The dataset is formatted as a single JSON file. Each record represents a sentence and contains the following fields:

• doc_id: Unique identifier of the sentence (contains the split name, e.g., wol_validation_x or wol_test_x).
• text: The raw Wolof sentence.
• tokens: List of words in the sentence.
• entities: A list of annotated entity mentions, where each entity contains:
  • start: Character start index in the sentence.
  • end: Character end index in the sentence.
  • text: The raw Wolof surface form of the mention.
  • ner_type: Named entity category (LOC, ORG, or PER).
  • token_start: Word token start index.
  • token_end: Word token end index.
  • qid: The correct Wikidata QID, or "NIL" if the entity is absent from Wikidata.
  • qid_label: The canonical Wikidata label (in French or English).
  • qid_url: Wikidata URL link to the entity.
  • qid_description_lang: The language code of the Wikidata description.
  • qid_description: Wikidata description of the entity.
  • source: Annotation source (always wikidata or nil).
  • annotation_status: The linking status (linked or nil).
  • annotation_method: The method used to annotate this mention (manual or semi-automatic).
  • nil_reason: The reason if the entity is NIL, or null.
  • nil_category: The category of NIL reason, or null.

JSON Example

{
  "doc_id": "wol_validation_2",
  "text": "Móritani ak Madagaskaar ñu ngi ci mbuus 3 gi .",
  "tokens": [
    "Móritani",
    "ak",
    "Madagaskaar",
    "ñu",
    "ngi",
    "ci",
    "mbuus",
    "3",
    "gi",
    "."
  ],
  "entities": [
    {
      "start": 0,
      "end": 8,
      "text": "Móritani",
      "ner_type": "LOC",
      "token_start": 0,
      "token_end": 1,
      "qid": "Q1025",
      "qid_label": "Mauritanie",
      "qid_url": "http://www.wikidata.org/entity/Q1025",
      "qid_description_lang": "fr",
      "qid_description": "pays d'Afrique de l'Ouest indépendant depuis 1960",
      "source": "wikidata",
      "annotation_status": "linked",
      "annotation_method": "manual",
      "nil_reason": null,
      "nil_category": null
    },
    {
      "start": 12,
      "end": 23,
      "text": "Madagaskaar",
      "ner_type": "LOC",
      "token_start": 2,
      "token_end": 3,
      "qid": "Q1019",
      "qid_label": "Madagascar",
      "qid_url": "http://www.wikidata.org/entity/Q1019",
      "qid_description_lang": "fr",
      "qid_description": "pays insulaire situé dans la partie occidentale de l'océan Indien",
      "source": "wikidata",
      "annotation_status": "linked",
      "annotation_method": "manual",
      "nil_reason": null,
      "nil_category": null
    }
  ]
}

Annotation Guidelines

Specific annotation guidelines were established to handle linguistic nuances:

1. Orthographic variations: Non-standardized names (e.g., Maki Sall / Macky Sall) are linked to the same unique QID.
2. Sports Metonymy: Country or city names referring to sports teams (e.g., Senegaal in football context) are resolved to the national team QID (Q206689) rather than the country QID (Q1041).
3. Geographical Granularity: Place names are linked to the physical settlement/village rather than administrative divisions unless context dictates otherwise.
4. Acronyms: Ambiguous local acronyms (e.g., SDE, ICS) are expanded and resolved to their correct regional organization entries.

Citation

If you use this dataset in your research, please cite it as:

@misc{mbaye2026wolofel,
  title={Wolof Entity Linking Dataset},
  author={Mbaye, Mouhamed},
  year={2026},
  publisher={Hugging Face},
  howpublished={\url{https://huggingface.co/datasets/m-mbaye30/WolofEntityLinking}}
}

Since the source texts originate from the MasakhaNER 2.0 dataset, please also cite the original work:

@article{Adelani2022MasakhaNER2A,
  title={MasakhaNER 2.0: Africa-centric Transfer Learning for Named Entity Recognition},
  author={David Ifeoluwa Adelani and Graham Neubig and Sebastian Ruder and Shruti Rijhwani and Michael Beukman and Chester Palen-Michel and Constantine Lignos and Jesujoba Oluwadara Alabi and Shamsuddeen Hassan Muhammad and Peter Nabende and Cheikh M. Bamba Dione and Andiswa Bukula and Rooweither Mabuya and Bonaventure F. P. Dossou and Blessing K. Sibanda and Happy Buzaaba and Jonathan Mukiibi and Godson Kalipe and Derguene Mbaye and Amelia Taylor and Fatoumata Kabore and Chris C. Emezue and Anuoluwapo Aremu and Perez Ogayo and Catherine W. Gitau and Edwin Munkoh-Buabeng and Victoire Memdjokam Koagne and Allahsera Auguste Tapo and Tebogo Macucwa and Vukosi Marivate and Elvis Mboning and Tajuddeen R. Gwadabe and Tosin P. Adewumi and Orevaoghene Ahia and Joyce Nakatumba-Nabende and Neo L. Mokono and Ignatius M Ezeani and Chiamaka Ijeoma Chukwuneke and Mofetoluwa Adeyemi and Gilles Hacheme and Idris Abdulmumin and Odunayo Ogundepo and Oreen Yousuf and Tatiana Moteu Ngoli and Dietrich Klakow},
  journal={ArXiv},
  year={2022},
  volume={abs/2210.12391}
}