DanishMahdi/snd-names

Dataset Description

The snd-names Dataset is a curated collection of Sindhi personal names designed for NLP and language modeling tasks in low-resource Sindhi language applications.

The dataset contains 10.3K unique rows of Sindhi names written in Arabic script. It is intended to support tasks such as named entity recognition (NER), text generation, OCR post-processing, and language modeling.

The names include a variety of common first names and culturally relevant Sindhi naming patterns.

---

language:
  - sd

task_categories:
  - named-entity-recognition
  - text-generation
  - language-modeling

task_ids:
  - token-classification
  - text-to-text-generation

splits:
  train: 10300

total_samples: 10300

---

Dataset Structure

features:
  name:
    type: string
    description: "Sindhi personal name in Unicode Arabic script"

Example:

name: "عبدالرحمن"

---

Intended Uses

suitable_for:
  - "Sindhi named entity recognition (NER)"
  - "Language model training and fine-tuning"
  - "OCR post-processing correction"
  - "Synthetic text generation"
  - "Linguistic research on Sindhi names"

not_recommended_for:
  - "Identity verification or real-world personal identification"
  - "Demographic or statistical population inference"

---

Limitations

limitations:
  - "The dataset contains only names and does not include contextual sentences."
  - "It may not represent all regional, tribal, or rare Sindhi naming variations."

---

Citation

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

@dataset{mahdi2026sindhinames,
  author    = {Danish Mahdi},
  title     = {Sindhi Names Dataset},
  year      = {2026},
  publisher = {Hugging Face},
  url       = {https://huggingface.co/datasets/DanishMahdi/snd-names}
}