README.md

Text Tagging Dataset

Overview

This dataset is designed for natural language processing (NLP) tasks, particularly for sequence tagging and named entity recognition (NER). It contains tokens labeled with specific tags to identify entities, categories, or relationships. The dataset is suitable for training and evaluating NER models, token classification, or other similar tasks.

---

Metadata Summary

Metadata	Value
Number of Sentences	1,472
Number of Unique Tags	42
Total Number of Tokens	9,528

---

Tag Frequency Distribution

The dataset uses a variety of tags to annotate tokens. Below is a summary of the tags and their occurrences:

Name of Tags	Count
O	8,194
B-gpe	207
B-per	168
B-num	149
B-org	111
I-org	90
B-dat	74
I-per	67
B-nat	59
B-tim	52
B-art	44
B-ani	40
B-eve	38
I-num	27
B-geo	27
I-gpe	25
I-eve	22
I-dat	13
E-org	13
B-bird	10

---

Structure of the Dataset

The dataset consists of the following columns:

• Sentence: Indicates the sentence to which the token belongs.
• Word: The token in the sentence.
• Tag: The tag assigned to the token, such as B-per, O, etc.

Sample Data

Sentence	Word	Tag
Sentence 1	تارا	B-per
Sentence 1	کتێبەکەی	O
Sentence 1	چاپ	O
Sentence 1	کرد	O
Sentence 2	مشک	B-ani

---

Usage

This dataset can be used for various NLP tasks, including:

1. Named Entity Recognition (NER): Training and evaluating models for entity extraction.
2. Token Classification: Identifying the roles or properties of tokens within a sequence.
3. Custom NLP Applications: Building systems requiring labeled text data.

---

Authors and Affiliations

• Abdulhady Abas Abdullah

  1. Artificial Intelligence and Innovation Centre, University of Kurdistan Hewler, Erbil, Iraq
     Email: [email protected]
  2. Computer Science and Engineering Department, University of Kurdistan Hewler, Erbil, KR, Iraq
     Email: [email protected]

• Srwa Hasan Abdulla

  3. Department of Horticulture, College of Agricultural Engineering Sciences, University of Sulaimani, Kurdistan Region, Iraq
     Email: [email protected]

• Dalia Mohammad Toufiq

---

NER-RoBERTa

NER-RoBERTa is a model fine-tuned for Named Entity Recognition (NER) in low-resource languages. This repository contains the model/dataset along with all necessary files and instructions for use.

---

Citation

If you use this model/dataset in your research or projects, please cite our work as follows:

@article{abdullah2024ner,
  title={NER-RoBERTa: Fine-Tuning RoBERTa for Named Entity Recognition (NER) within low-resource languages},
  author={Abdullah, Abdulhady Abas and Abdulla, Srwa Hasan and Toufiq, Dalia Mohammad and Maghdid, Halgurd S and Rashid, Tarik A and Farho, Pakshan F and Sabr, Shadan Sh and Taher, Akar H and Hamad, Darya S and Veisi, Hadi and others},
  journal={arXiv preprint arXiv:2412.15252},
  year={2024}
}