Model Card for Arabic Named Entity Recognition with AraBERT
Model Details
Model Name: AraBERT-NER
Model Type: AraBERT (Pre-trained on Arabic text and fine-tuned on Arabic Named Entity Recognition task)
Language: Arabic
License: MIT
Model Creator: Mostafa Ahmed
Contact Information: mostafa.ahmed00976@gmail.com
Model Version: 1.0
Overview
AraBERT-NER is a fine-tuned version of the AraBERT model specifically designed for Named Entity Recognition (NER) tasks in Arabic. The model has been trained to identify and classify named entities such as persons, organizations, locations and MISC and more within Arabic text. This makes it suitable for various applications such as information extraction, document categorization, and data annotation in Arabic.
Intended Use
The model is intended for use in:
- Named Entity Recognition systems for Arabic
- Information extraction from Arabic text
- Document categorization and annotation
- Arabic language processing research
Training Data
The model was fine-tuned on the CoNLL-NER-AR dataset.
Data Sources:
- CoNLL-NER-AR: A dataset for named entity recognition tasks in Arabic.
Training Procedure
The model was trained using the Hugging Face transformers library. The training process involved:
- Preprocessing the CoNLL-NER-AR to format the text and entity annotations for NER.
- Fine-tuning the pre-trained AraBERT model on the Arabic NER dataset.
- Evaluating the model's performance using standard NER metrics (e.g., Precision, Recall, F1 Score).
Evaluation Results
The model was evaluated on a held-out test set from the CoNLL-NER-AR dataset. Here are the key performance metrics:
- Precision: 0.8547
- Recall: 0.8633
- F1 Score: 0.8590
- Accuracy: 0.9542
These metrics indicate the model's effectiveness in accurately identifying and classifying named entities in Arabic text.
How to Use
You can load and use the model with the Hugging Face transformers library as follows:
from transformers import AutoTokenizer, AutoModelForTokenClassification
from transformers import pipeline
tokenizer = AutoTokenizer.from_pretrained("MostafaAhmed98/AraBert-Arabic-NER-CoNLLpp")
model = AutoModelForTokenClassification.from_pretrained("MostafaAhmed98/AraBert-Arabic-NER-CoNLLpp")
# Create a NER pipeline
ner_pipeline = pipeline("ner", model=model, tokenizer=tokenizer)
# Example usage
text = "ولد محمد علي في القاهرة وعمل في شركة مايكروسوفت."
ner_results = ner_pipeline(text)
for entity in ner_results:
print(f"Entity: {entity['word']}, Label: {entity['entity']}, Confidence: {entity['score']:.2f}")