Skip to content

A simple and interactive Streamlit web app to classify Indonesian text sentiment (positive, negative, or neutral) using IndoBERT, a pre-trained BERT model fine-tuned for sentiment analysis.

Notifications You must be signed in to change notification settings

Adittt17/sentiment_analysis_indonlu

Folders and files

NameName
Last commit message
Last commit date

Latest commit

 

History

7 Commits
 
 
 
 
 
 
 
 
 
 
 
 

Repository files navigation

🇮🇩 IndoBERT Sentiment Classifier

An accurate and efficient sentiment classification system for Indonesian text, powered by IndoBERT. This project demonstrates how a fine-tuned transformer model can effectively classify sentiment in real-world Indonesian documents.

🧠 Model Highlights

  • Based on indobenchmark/indobert-base-p1 — a pre-trained BERT model for the Indonesian language
  • Fine-tuned on document-level sentiment dataset
  • Supports 3 sentiment classes:
    • Negative
    • Neutral
    • Positive
  • Achieves strong performance on test data with high precision and recall
  • Inference-ready and optimized for deployment

📌 Use Case

This model is designed for Indonesian-language applications such as:

  • Social media monitoring
  • Customer feedback analysis
  • Product review classification
  • Public opinion mining

🔍 Example

Input:

"Merasa kagum dengan toko ini tapi berubah menjadi kecewa setelah transaksi"

Output:

🧠 Prediction: negative (93.2%)

⚙️ Tech Stack

  • PyTorch — deep learning framework
  • HuggingFace Transformers — for loading and managing the IndoBERT model
  • Git LFS — to store large model weights (>500MB)
  • Streamlit — for quick demo deployment (optional)

📺 Try the Web App

Want to see the model in action?
👉 Access the live UI here: https://sentiment-analysis-indonlu.streamlit.app/

📁 Model File

Make sure the trained model is stored in the following path:

  • model/best_model.pt

⚠️ Note: The model file is tracked using Git LFS due to its size.

🧪 Training

The model was trained using a custom implementation of the DocumentSentimentDataset and DocumentSentimentDataLoader from IndoNLU. Training utilized:

  • Adam optimizer
  • Custom metrics calculation
  • GPU acceleration (CUDA)
  • Validation-based evaluation per epoch

📚 Acknowledgements

  • IndoNLU — for Indonesian NLP datasets and benchmarks
  • HuggingFace — for providing model architectures and tokenizer support
  • Adityo Pangestu — for training, optimizing, and deploying the model

📬 Contact

Created by Adityo Pangestu · adityopangestu01@gmail.com
Feel free to contribute or extend this project for other NLP tasks such as topic modeling, emotion detection, or intent classification.

About

A simple and interactive Streamlit web app to classify Indonesian text sentiment (positive, negative, or neutral) using IndoBERT, a pre-trained BERT model fine-tuned for sentiment analysis.

Topics

Resources

Stars

Watchers

Forks

Releases

No releases published

Packages

No packages published
pFad - Phonifier reborn

Pfad - The Proxy pFad of © 2024 Garber Painting. All rights reserved.

Note: This service is not intended for secure transactions such as banking, social media, email, or purchasing. Use at your own risk. We assume no liability whatsoever for broken pages.


Alternative Proxies:

Alternative Proxy

pFad Proxy

pFad v3 Proxy

pFad v4 Proxy