README.md

metadata

title: Sentiment Analysis API
emoji: 😊
colorFrom: blue
colorTo: green
sdk: docker
pinned: false
license: mit

Sentiment Analysis API

A FastAPI-based sentiment analysis service that predicts sentiment (positive/negative) from text input.

Features

• Sentiment Prediction: Get integer predictions (0 for negative, 1 for positive)
• Probability Scores: Get prediction probabilities for both classes
• Batch Processing: Analyze multiple texts at once
• Interactive API: Swagger UI documentation available at /docs

API Endpoints

/predict

• Method: POST
• Input: JSON with text field
• Output: Prediction, confidence score, and sentiment label

/predict_proba

• Method: POST
• Input: JSON with text field
• Output: Probability array, prediction, and sentiment label

/batch_predict

• Method: POST
• Input: Array of text strings
• Output: Results for all input texts

Usage Example

import requests

# Single prediction
response = requests.post(
    "https://your-space-url/predict",
    json={"text": "I love this movie!"}
)
print(response.json())
# Output: {"prediction": 1, "confidence": 0.95, "sentiment": "positive"}

# Probability prediction
response = requests.post(
    "https://your-space-url/predict_proba", 
    json={"text": "This is terrible"}
)
print(response.json())
# Output: {"probabilities": [0.85, 0.15], "prediction": 0, "sentiment": "negative"}

Local Development

1. Install dependencies: pip install -r requirements.txt
2. Run the app: python app.py
3. Visit http://localhost:7860/docs for interactive API documentation