README.md

metadata

license: cc-by-4.0
task_categories:
  - text-classification
task_ids:
  - multi-class-classification
language:
  - ckb
multilinguality:
  - monolingual
size_categories:
  - 1K<n<10K
pretty_name: Kurdish Poetry Meter Classification
tags:
  - poetry
  - kurdish
  - central-kurdish
  - meter-classification
  - prosody
  - multi-class-classification
  - mteb
  - poetrymteb
  - embedding-evaluation
annotations_creators:
  - expert-generated
source_datasets:
  - AsoSoft/Vejinbooks-Poem-Dataset
configs:
  - config_name: default
    data_files:
      - split: train
        path: data/train-*
      - split: validation
        path: data/validation-*
      - split: test
        path: data/test-*
    default: true
dataset_info:
  - config_name: default
    features:
      - name: id
        dtype: string
      - name: poet
        dtype: string
      - name: poem_number
        dtype: string
      - name: title
        dtype: string
      - name: form
        dtype: string
      - name: poem
        dtype: string
      - name: label
        dtype: int64
      - name: label_name
        dtype: string
    splits:
      - name: train
        num_examples: 624
      - name: validation
        num_examples: 208
      - name: test
        num_examples: 209

Kurdish Poetry Meter Classification

Multi-class meter (prosody) classification of Central Kurdish poetry for PoetryMTEB embedding evaluation.

Derived from the VejinBooks Poem Dataset (AsoSoft): expert-annotated meter and form tags for poems from vejinbooks. Used in automatic Kurdish meter identification research (Mahmudi & Veisi, PLOS ONE 2023).

Dataset Card

Item	Description
Dataset version (PoetryMTEB)	1.0.0
Source (GitHub)	AsoSoft/Vejinbooks-Poem-Dataset
Source (Zenodo)	https://doi.org/10.5281/zenodo.4079471
Paper	Mahmudi A, Veisi H. Automatic meter classification of Kurdish poems. PLOS ONE, 2023 (DOI)
Languages	Central Kurdish (ckb)
Unit	Full poem text (poem)
Labels	Single-label 15 meter patterns (label / label_name)
Size	train=624; validation=208; test=209; total=1041
Splits	Preserved from the previous PoetryMTEB release (~6:2:2 stratified by meter)
License	CC BY 4.0 (PoetryMTEB redistribution; cite original authors)
Evaluation metrics	Embedding classification: accuracy, macro/micro F1 (see scripts/evaluate_kurdish_meter.py)

Note on size: Upstream VejinBooks release has 1154 poems. This packaging keeps 1041 poems after filtering to meters with ≥10 examples (and format checks), matching the prior PoetryMTEB release.

Label taxonomy (15 meters)

id	label_name (meter)	train	validation	test	total
0	مفاعیلن مفاعیلن مفاعیلن مفاعیلن	146	48	49	243
1	فاعلاتن فاعلاتن فاعلاتن فاعلن	132	44	44	220
2	مفاعیلن مفاعیلن فعولن	79	26	27	132
3	مفعولُ مفاعیلُ مفاعیلُ فعولن	69	23	24	116
4	فعلاتن فعلاتن فعلاتن فعلن	44	14	15	73
5	10Syllabic	27	9	10	46
6	FreeVerse	26	9	8	43
7	فاعلاتن فاعلاتن فاعلن	20	7	6	33
8	مفعولُ فاعلاتُ مفاعیلُ فاعلن	18	7	6	31
9	8Syllabic	18	6	6	30
10	فعلاتن مفاعلن فعلن	15	5	5	25
11	مفعولُ مفاعیلن مفعولُ مفاعیلن	10	3	3	16
12	فعلاتن فعلاتن فعلن	8	3	2	13
13	7Syllabic	6	2	2	10
14	مفعولُ فاعلاتن مفعولُ فاعلاتن	6	2	2	10

Codebook: label_taxonomy.json.

Features

Field	Type	Description
id	string	Example id (kpmc-{split}-{legacy_id})
poet	string	Poet name (from upstream filename, e.g. Diɫdar)
poem_number	string	Poem number within poet (e.g. 002)
title	string	Poem title (often first hemistich)
form	string	Poetic form in Kurdish (e.g. غەزەل, مەسنەوی, نوێ)
poem	string	Classification input: poem body (# couplet markers removed; newlines kept)
label	int64	Meter class id
label_name	string	Meter pattern string (formerly meter)

Schema aligns with other PoetryMTEB classification datasets (id, poem, label, label_name).

Construction method

1. Load existing PoetryMTEB/KurdishPoetryMeterClassification parquet splits.
2. Normalize fields: text→poem, meter→label_name + integer label.
3. Keep poet, poem_number, title, form; assign stable ids.
4. Preserve train / validation / test membership (no re-split in this packaging pass).
5. Write parquet + dataset_infos.yaml + Dataset Card + supporting materials.

Upstream file format (reference)

title: ...
form: ...
meter: ...
<poem text from line 4>

• # marks a new couplet (stripped in this packaging).
• ++ marks non-Kurdish hemistiches in upstream files.

Supporting materials (this repo)

Path	Content
VERSION	PoetryMTEB packaging version
LICENSE	CC BY 4.0 notice
CITATION.cff / CITATION.bib	Citation metadata
metadata.json	Dataset metadata
label_taxonomy.json	Meter codebook + counts
splits/*_ids.txt	Split legacy id lists
checksums.sha256	SHA256 of packaged files
prompts/annotation_prompt_template.md	Meter labeling prompt template
scripts/evaluate_kurdish_meter.py	Baseline probe evaluation

How to load

from datasets import load_dataset

ds = load_dataset("PoetryMTEB/KurdishPoetryMeterClassification")
print(ds["train"][0]["poem"][:120])
print(ds["train"][0]["label_name"], ds["train"][0]["label"])

Intended use

• PoetryMTEB / MTEB-style multi-class classification probing of Central Kurdish poem embeddings (meter / prosody recognition).
• Research on Kurdish quantitative, syllabic, and free-verse meters.

Citation

@article{mahmudi2023meter,
  author  = {Mahmudi, Aso and Veisi, Hadi},
  title   = {Automatic meter classification of Kurdish poems},
  journal = {PLOS ONE},
  year    = {2023},
  doi     = {10.1371/journal.pone.0280263},
  url     = {https://doi.org/10.1371/journal.pone.0280263}
}

Please also cite the dataset release:

@dataset{vejinbooks_poem_dataset,
  author = {Mahmudi, Aso},
  title  = {VejinBooks Poem Dataset},
  year   = {2020},
  doi    = {10.5281/zenodo.4079471},
  url    = {https://github.com/AsoSoft/Vejinbooks-Poem-Dataset}
}

Upstream: https://github.com/AsoSoft/Vejinbooks-Poem-Dataset · Zenodo: https://doi.org/10.5281/zenodo.4079471
This Hub packaging: PoetryMTEB/KurdishPoetryMeterClassification (version 1.0.0)

License

Distributed under CC BY 4.0 for this PoetryMTEB packaging. Please cite the PLOS ONE paper and Zenodo/GitHub dataset when using the data.