README.md

metadata

library_name: transformers
pipeline_tag: token-classification
tags:
  - named-entity-recognition
  - merchant-extraction
  - finance
  - pytorch
language:
  - en

Merchant Name Extraction Model

This model extracts merchant names from transaction descriptions using Named Entity Recognition (NER).

Model Details

• Model Type: DistilBERT for Token Classification
• Task: Merchant Name Extraction
• Language: English
• Framework: PyTorch + Transformers

Usage

from transformers import DistilBertTokenizerFast, DistilBertForTokenClassification
import torch

# Load model and tokenizer
model = DistilBertForTokenClassification.from_pretrained("GalalEwida/SIA-MerchentName")
tokenizer = DistilBertTokenizerFast.from_pretrained("GalalEwida/SIA-MerchentName")

# Prediction function
def extract_merchant(text):
    inputs = tokenizer(text, return_tensors="pt", truncation=True, max_length=128)
    
    with torch.no_grad():
        outputs = model(**inputs)
        predictions = torch.argmax(outputs.logits, dim=2)
    
    tokens = tokenizer.convert_ids_to_tokens(inputs['input_ids'][0])
    id2label = {0: 'O', 1: 'B-MERCHANT', 2: 'I-MERCHANT'}
    predicted_labels = [id2label[pred.item()] for pred in predictions[0]]
    
    merchant_tokens = []
    for token, label in zip(tokens, predicted_labels):
        if label in ['B-MERCHANT', 'I-MERCHANT']:
            if token.startswith('##'):
                if merchant_tokens:
                    merchant_tokens[-1] += token[2:]
            else:
                merchant_tokens.append(token)
    
    return ' '.join(merchant_tokens)

# Example usage
text = "WALMART SUPERCENTER #1234 ANYTOWN US"
merchant = extract_merchant(text)
print(f"Extracted: {merchant}")

Labels

• O: Outside (not part of merchant name)
• B-MERCHANT: Beginning of merchant name
• I-MERCHANT: Inside merchant name

Example Predictions

Input	Extracted Merchant
WALMART SUPERCENTER #1234 ANYTOWN US	WALMART
AMAZON.COM AMZN.COM/BILL WA	AMAZON
STARBUCKS STORE #0123 NEW YORK NY	STARBUCKS