Flair-Persian-NER
| Property | Value |
|---|---|
| Author | PooryaPiroozfar |
| Framework | Flair, PyTorch |
| Dataset | NSURL-2019 |
| Performance | 90.33% F1-Score |
| Downloads | 3,501 |
What is Flair-Persian-NER?
Flair-Persian-NER is a specialized Named Entity Recognition model designed for the Persian language. Built on the Flair framework and leveraging Pars-Bert embeddings, this model can identify and classify seven different types of named entities in Persian text with high accuracy. The model represents a significant advancement in Persian language processing capabilities.
Implementation Details
The model implements a sequence tagging architecture based on Flair embeddings and Pars-Bert. It achieves impressive performance metrics with a micro F1-score of 90.33% and macro F1-score of 89.76% on the NSURL-2019 dataset. The implementation supports real-time entity recognition and can be easily integrated using the Flair library.
- Supports 7 entity types: Person (PER), Location (LOC), Organization (ORG), Date (DAT), Time (TIM), Percent (PCT), and Money (MON)
- Built on state-of-the-art Flair embeddings and Pars-Bert architecture
- Achieves high precision across all entity types, particularly strong in Money (96.65%) and Percent (93.75%) recognition
Core Capabilities
- High-accuracy entity recognition with 90.33% overall F1-score
- Specialized Persian language support
- Simple integration through Flair library
- Real-time text analysis capabilities
- Comprehensive entity type coverage
Frequently Asked Questions
Q: What makes this model unique?
The model's specialization in Persian language NER, combined with its high accuracy and comprehensive entity coverage, makes it particularly valuable for Persian text analysis. Its integration with the popular Flair framework ensures easy deployment and usage.
Q: What are the recommended use cases?
The model is ideal for applications requiring Persian text analysis, including information extraction, document processing, automated content tagging, and business intelligence systems working with Persian language content. It's particularly effective for identifying and classifying names, locations, organizations, dates, times, and numerical expressions.
