Datasets:

agentlans
/

wikipedia-first-paragraph-ner

configs:
  - config_name: en
    data_files:
      - path:
          - en.jsonl.zst
        split: train
    default: true
  - config_name: zh
    data_files:
      - path:
          - zh.jsonl.zst
        split: train
  - config_name: ja
    data_files:
      - path:
          - ja.jsonl.zst
        split: train
license: cc-by-sa-4.0
task_categories:
  - token-classification
  - feature-extraction
language:
  - en
  - zh
  - ja
tags:
  - ner
  - wikipedia
  - spacy

Wikipedia First Paragraph NER

This dataset contains named entities extracted from the first paragraph of every article in the agentlans/wikipedia-first-paragraph Wikipedia dataset. Entity recognition was performed using spaCy’s en_core_web_sm, zh_core_web_sm, ja_core_news_sm models.

Each record corresponds to one Wikipedia article and includes the article title along with lists of entities grouped by their type.

Example:

{
  "title": "Moot Hall, Brampton",
  "CARDINAL": [],
  "DATE": [],
  "EVENT": [],
  "FAC": [],
  "GPE": ["Brampton", "England", "Cumbria"],
  "LANGUAGE": [],
  "LAW": [],
  "LOC": [],
  "MONEY": [],
  "NORP": [],
  "ORDINAL": [],
  "ORG": ["Brampton Parish Council", "The Market Place"],
  "PERCENT": [],
  "PERSON": ["The Moot Hall"],
  "PRODUCT": [],
  "QUANTITY": [],
  "TIME": [],
  "WORK_OF_ART": []
}

Field Descriptions

Field	Description
`title`	Title of the Wikipedia article.
`CARDINAL`	Numerals that do not fall under another type.
`DATE`	Absolute or relative dates or time periods.
`EVENT`	Named events such as wars, battles, or sports competitions.
`FAC`	Facilities such as buildings, airports, highways, or bridges.
`GPE`	Geopolitical entities: countries, cities, states.
`LANGUAGE`	Named languages.
`LAW`	Named legal documents or statutes.
`LOC`	Non-GPE locations, including mountains and bodies of water.
`MONEY`	Monetary values including currency units.
`NORP`	Nationalities, religious groups, or political groups.
`ORDINAL`	Ordinal numbers (e.g., “first”, “second”).
`ORG`	Organizations, companies, institutions, or agencies.
`PERCENT`	Percent expressions including “%”.
`PERSON`	Individual people, including fictional characters.
`PRODUCT`	Physical products such as objects, vehicles, or foods.
`QUANTITY`	Measured quantities (e.g., weight, distance).
`TIME`	Specific times smaller than a day.
`WORK_OF_ART`	Creative works such as book, song, or film titles.

Limitations

Many Wikipedia articles are short or stubs, resulting in limited content for entity extraction.
spaCy’s models may produce both false positives and false negatives.
Misclassifications are possible (for example, botanical names detected as PERSON, or person names as ORG).
Data originates from a static Wikipedia dump and may not reflect the latest content.

Licence

Creative Commons Attribution-ShareAlike 4.0 International (CC BY-SA 4.0)