niejanee/tokopedia-sentiment-analysis-indobert

Indonesian Sentiment Analysis with IndoBERT

This model is a fine-tuned version of indobenchmark/indobert-base-p2 for sentiment analysis of Indonesian customer reviews.

Model Description

• Model Type: BERT-based sequence classification
• Language: Indonesian (Bahasa Indonesia)
• Base Model: indobenchmark/indobert-base-p2
• Task: Sentiment Analysis (3 classes: Negative, Neutral, Positive)
• Dataset: Tokopedia Product Reviews 2019

Intended Uses

This model is intended for sentiment analysis of Indonesian text, particularly customer reviews and product feedback.

Direct Use

from transformers import pipeline

# Load the model
sentiment_pipeline = pipeline("text-classification", model="niejanee/tokopedia-sentiment-analysis-indobert")

# Analyze sentiment
result = sentiment_pipeline("Produk sangat memuaskan, kualitas premium!")
print(result)
# Output: [{'label': 'LABEL_2', 'score': 0.9876}]  # LABEL_2 = Positive

API Usage

import requests

API_URL = "https://api-inference.huggingface.co/models/niejanee/tokopedia-sentiment-analysis-indobert"
headers = {"Authorization": "Bearer YOUR_HF_TOKEN"}

def query(payload):
    response = requests.post(API_URL, headers=headers, json=payload)
    return response.json()

output = query({
    "inputs": "Produk sangat bagus!",
})
print(output)

Label Mapping

• LABEL_0: Negative sentiment
• LABEL_1: Neutral sentiment
• LABEL_2: Positive sentiment

Training Details

Training Data

The model was trained on Indonesian customer reviews from Tokopedia with the following distribution:

• Negative: Reviews with ratings 1-2
• Neutral: Reviews with rating 3
• Positive: Reviews with ratings 4-5

Training Procedure

• Training Framework: Hugging Face Transformers
• Optimizer: AdamW
• Learning Rate: 2e-5
• Batch Size: 16
• Epochs: 3
• Max Sequence Length: 512

Usage Examples

Python

from transformers import AutoTokenizer, AutoModelForSequenceClassification, pipeline

# Load model and tokenizer
model = AutoModelForSequenceClassification.from_pretrained("niejanee/tokopedia-sentiment-analysis-indobert")
tokenizer = AutoTokenizer.from_pretrained("niejanee/tokopedia-sentiment-analysis-indobert")

# Create pipeline
classifier = pipeline("text-classification", model=model, tokenizer=tokenizer)

# Predict sentiment
texts = [
    "Barang sangat bagus, pelayanan memuaskan!",
    "Produk tidak sesuai ekspektasi",
    "Kualitas standar, harga wajar"
]

results = classifier(texts)
for text, result in zip(texts, results):
    print(f"Text: {text}")
    print(f"Sentiment: {result['label']} (Score: {result['score']:.4f})")

JavaScript

async function analyzeSentiment(text) {
    const response = await fetch(
        "https://api-inference.huggingface.co/models/niejanee/tokopedia-sentiment-analysis-indobert",
        {
            method: "POST",
            headers: {
                "Authorization": "Bearer YOUR_HF_TOKEN",
                "Content-Type": "application/json",
            },
            body: JSON.stringify({
                "inputs": text
            }),
        }
    );
    
    const result = await response.json();
    return result;
}

Limitations

• The model is specifically trained on Indonesian customer reviews and may not perform well on other types of Indonesian text
• Performance may vary on informal language, slang, or text with many typos
• The model was trained on e-commerce reviews and may have domain-specific biases

License

This model is licensed under Apache 2.0.