bert-fa-base-uncased-ner-arman Open Source Model - Optimize Persian Language Understanding and Expand Application Functions in Multiple Fields

Bert Fa Base Uncased Ner Arman

Developed by HooshvareLab

A Transformer-based Persian language understanding model, reconstructed vocabulary and fine-tuned on new corpora, expanding multi-domain application capabilities

Sequence Labeling OtherOpen Source License:Apache-2.0 #Persian Named Entity Recognition #Multi-domain Adaptation #High-precision NER

Downloads 110

Release Time : 3/2/2022

Model Overview

ParsBERT is a BERT model optimized for Persian, primarily used for natural language understanding tasks such as Named Entity Recognition (NER). Version 2.0 improves model performance and application scope by reconstructing the vocabulary and fine-tuning on new corpora.

Model Features

Optimized Persian Processing

Specifically reconstructed vocabulary for Persian to enhance language understanding

Multi-domain Applicability

Fine-tuned on new corpora, expanding the model's application across different domains

High-performance NER

Achieves an F1 score of 99.84 in Persian Named Entity Recognition tasks

Model Capabilities

Persian text understanding

Named Entity Recognition

Token classification

Use Cases

Information Extraction

Persian Text Entity Recognition

Identify entities such as organizations, locations, and people from Persian text

Achieved an F1 score of 99.84 on the ARMAN dataset

🚀 ParsBERT (v2.0)

A Transformer-based Model for Persian Language Understanding

We reconstructed the vocabulary and fine-tuned ParsBERT v1.1 on the new Persian corpora to offer additional functionalities for using ParsBERT in other domains. For the latest information on previous and current models, please refer to the ParsBERT repository.

✨ Features

Persian NER [ARMAN, PEYMA]

This task focuses on extracting named entities from text, such as names, and labeling them with appropriate NER classes like locations and organizations. The datasets used for this task contain sentences marked in the IOB format. In this format, tokens not part of an entity are tagged as "O", the "B" tag indicates the first word of an object, and the "I" tag represents the remaining terms of the same entity. Both "B" and "I" tags are followed by a hyphen (or underscore), then the entity category. Thus, the NER task is a multi-class token classification problem that labels tokens given raw text. There are two main datasets used in Persian NER: ARMAN and PEYMA.

ARMAN

The ARMAN dataset consists of 7,682 sentences, with 250,015 tokens tagged across six different classes:

Organization
Location
Facility
Event
Product
Person

Label	#
Organization	30108
Location	12924
Facility	4458
Event	7557
Product	4389
Person	15645

Download You can download the dataset from here.

📚 Documentation

Results

The following table summarizes the F1 score achieved by ParsBERT compared to other models and architectures.

Dataset	ParsBERT v2	ParsBERT v1	mBERT	MorphoBERT	Beheshti-NER	LSTM-CRF	Rule-Based CRF	BiLSTM-CRF
ARMAN	99.84*	98.79	95.89	89.9	84.03	86.55	-	77.45

💻 Usage Examples

Notebook	Description
How to use Pipelines	A simple and efficient way to use State-of-the-Art models on downstream tasks through transformers

BibTeX entry and citation info

Please cite in publications as the following:

@article{ParsBERT,
    title={ParsBERT: Transformer-based Model for Persian Language Understanding},
    author={Mehrdad Farahani, Mohammad Gharachorloo, Marzieh Farahani, Mohammad Manthouri},
    journal={ArXiv},
    year={2020},
    volume={abs/2005.12515}
}

📄 License

This project is licensed under the Apache-2.0 license.

❓ Questions

Post a Github issue on the ParsBERT Issues repository.

Featured Recommended AI Models

Empowering the Future, Your AI Solution Knowledge Base

English 简体中文繁體中文にほんご