metadata

license: mit
task_categories:
  - token-classification
language:
  - fa
pretty_name: PEYMA-ARMAN-Mixed
size_categories:
  - 10K<n<100K

Mixed Persian NER Dataset (PEYMA-ARMAN)

This dataset is a combination of PEYMA and ARMAN Persian NER datasets. It contains the following named entity tags:

Product (PRO)
Event (EVE)
Facility (FAC)
Location (LOC)
Person (PER)
Money (MON)
Percent (PCT)
Date (DAT)
Organization (ORG)
Time (TIM)

Dataset Information

The dataset is divided into three splits: train, test, and validation. Below is a summary of the dataset statistics:

Split	B_DAT	B_EVE	B_FAC	B_LOC	B_MON	B_ORG	B_PCT	B_PER	B_PRO	B_TIM	I_DAT	I_EVE	I_FAC	I_LOC	I_MON	I_ORG	I_PCT	I_PER	I_PRO	I_TIM	O	num_rows
Train	1512	1379	1334	13040	446	15762	266	11371	1719	224	1939	4600	2222	4254	1314	21347	308	7160	1736	375	747216	26417
Test	185	218	124	1868	53	2017	27	1566	281	27	245	697	237	511	142	2843	31	1075	345	37	92214	3303
Validation	161	143	192	1539	28	2180	33	1335	172	30	217	520	349	494	54	2923	34	813	136	39	96857	3302

First schema

DatasetDict({
    train: Dataset({
        features: ['tokens', 'ner_tags', 'ner_tags_names'],
        num_rows: 26417
    })
    test: Dataset({
        features: ['tokens', 'ner_tags', 'ner_tags_names'],
        num_rows: 3303
    })
    validation: Dataset({
        features: ['tokens', 'ner_tags', 'ner_tags_names'],
        num_rows: 3302
    })
})

How to load datset

from datasets import load_dataset
data = load_dataset("AliFartout/PEYMA-ARMAN-Mixed")

Feel free to adjust the formatting according to your needs.