Datasets:

hamidsalimi
/

Persian-NER-Dataset

license: mit
language:
  - fa
tags:
  - ner
  - named-entity-recognition
  - persian
  - farsi
  - token-classification
task_categories:
  - token-classification
pretty_name: Persian NER Dataset (Synthetic)

Persian NER Dataset (Synthetic)

Dataset Description

This dataset is a synthetically generated Named Entity Recognition (NER) dataset for the Persian (Farsi) language. It covers a wide range of entity types commonly found in Persian news articles, official documents, and general text.

Entities Covered

The dataset includes annotations for 23 distinct entity types:

LANGUAGE - Languages (e.g., فارسی, عربی)
LOCATION - Locations (e.g., خلیج فارس)
CITY - Cities (e.g., تهران, شیراز)
PROVINCE - Provinces (e.g., سیستان و بلوچستان)
FACILITY - Facilities & landmarks (e.g., برج میلاد)
ORGANIZATION - Organizations (e.g., سازمان ملل متحد)
PERSON - People (e.g., حسن روحانی)
EVENT - Events (e.g., انتخابات ریاست جمهوری)
DATE - Dates (e.g., ۱۴۰۳/۰۵/۱۰)
TIME - Times (e.g., ساعت ۱۴:۳۰)
NARCOTICS - Drugs (e.g., ماری‌جوانا)
WEAPON - Weapons (e.g., کلاشنیکف)
ALCOHOLIC_BEVERAGES - Alcohol (e.g., ودکا)
PRODUCT - Products (e.g., سمند)
MONEY - Money amounts (e.g., ۱۰۰ میلیون تومان)
PERCENT - Percentages (e.g., ۵۰ درصد)
POLITICAL_PARTY - Political parties (e.g., جبهه پایداری)
RELIGION - Religions (e.g., شیعه)
NATIONALITY - Nationalities (e.g., ایرانی)
LAW - Laws (e.g., قانون اساسی)
CULTURAL_CONCEPT - Cultural concepts (e.g., نوروز)
COUNTY - Counties (e.g., بخش ماهان)
WEAPON - Weapons (includes various weapon types)

Dataset Structure

Each example contains:

tokenized_text: A list of tokens
ner: Annotations with start/end indices and entity types

Usage

from datasets import load_dataset

dataset = load_dataset("your-username/persian-ner")