serendip-travel-classifier

Serendip Travel Experiential Classifier

A fine-tuned bert-base-uncased model for classifying Sri Lankan tourist reviews into four experiential dimensions.

Model Description

This model is a fine-tuned bert-base-uncased model trained on Sri Lankan tourism reviews to classify text into four experiential dimensions:

1. Regenerative & Eco-Tourism: Travel focused on positive social and environmental impact
2. Integrated Wellness: Journeys combining physical and mental well-being
3. Immersive Culinary: Experiences centered on authentic local cuisine
4. Off-the-Beaten-Path Adventure: Exploration of less crowded natural landscapes

Performance

• F1-Score: 0.9250
• Accuracy: 0.9233
• Precision: 0.9681
• Recall: 0.8859

Training Details

• Model: bert-base-uncased
• Batch Size: 8
• Learning Rate: 2e-05
• Epochs: 5

Usage

from transformers import AutoTokenizer, AutoModelForSequenceClassification
import torch

# Load model and tokenizer
tokenizer = AutoTokenizer.from_pretrained("j2damax/serendip-travel-classifier")
model = AutoModelForSequenceClassification.from_pretrained("j2damax/serendip-travel-classifier")

# Example text
text = "The organic tea plantation tour was amazing! We learned about sustainable farming practices."

# Tokenize and predict
inputs = tokenizer(text, return_tensors="pt", truncation=True, padding=True)
with torch.no_grad():
    outputs = model(**inputs)
    predictions = torch.sigmoid(outputs.logits)

# Get predicted labels
labels = ["Regenerative & Eco-Tourism", "Integrated Wellness", "Immersive Culinary", "Off-the-Beaten-Path Adventure"]
predicted_labels = [labels[i] for i, score in enumerate(predictions[0]) if score > 0.5]
print(f"Predicted labels: predicted_labels")

Model Card

This model was trained on Sri Lankan tourism reviews to classify experiences into four categories. The model uses a multi-label classification approach, meaning a single review can be classified into multiple categories simultaneously.

Limitations

• Trained specifically on Sri Lankan tourism data
• May not generalize well to other geographical regions
• Performance may vary with different text styles or languages

Citation

If you use this model, please cite:

@misc{serendip-travel-classifier,
  title={Serendip Travel Experiential Classifier},
  author={Jayampathy Balasuriya},
  year={2025},
  publisher={Hugging Face},
  url={https://huggingface.co/j2damax/serendip-travel-classifier}
}