README.md

---
title: NLP Sentiment Analysis App
emoji: 🚀
colorFrom: blue
colorTo: purple
sdk: gradio
sdk_version: "3.50.0"
app_file: app.py
pinned: false
---

# NLP Sentiment Analysis 🚀

This is a **sentiment analysis web app** built using **Hugging Face Transformers** and **Gradio**, deployed on **Hugging Face Spaces**.

## 🌟 Features
- ✅ Uses **DistilBERT** for sentiment analysis.
- ✅ Provides **positive/negative** sentiment classification.
- ✅ **Web-based UI** using Gradio.
- ✅ **Easy deployment** on Hugging Face Spaces.

---



# Personalized Mental Health Assistant

## Overview
This project is an AI-powered **Mental Health Assistant** that analyzes user queries, detects emotions, and provides relevant responses and resources. It leverages **NLP (Natural Language Processing)** techniques and **Machine Learning (ML)** models for sentiment analysis, emotion classification, and response generation.

## Features
- **Sentiment & Emotion Detection**: Classifies user queries based on emotions (Happy, Sad, Anxious, Overwhelmed, etc.).
- **Preprocessing Pipeline**: Cleans and preprocesses text data using **NLTK**.
- **TF-IDF Vectorization**: Converts user queries into numerical representations.
- **ML Classification**: Uses a **Random Forest Classifier** to predict emotions.
- **Response Generation**: Suggests appropriate responses based on detected emotions.
- **AWS Deployment**: Deploys as an API using **Flask & AWS Lambda**.

## Dataset Columns
The dataset includes the following columns:
| Column | Description |
|--------|-------------|
| ID | Unique identifier for each query |
| User Query | Raw text input from the user |
| Preprocessed Text | Cleaned version of the user query |
| Sentiment | Sentiment label (Positive, Neutral, Negative) |
| Emotion | Emotion label (e.g., Happy, Sad, Anxious) |
| Severity | Severity level of distress |
| Intent | Intent classification for responses |
| Suggested Response | AI-generated response for user support |
| Recommended Action | Suggested action based on the query |
| Related Resources | External resources for mental health support |
| Time of Interaction | Timestamp of the query |

---
## Installation
### **1. Clone Repository**
```bash
git clone https://github.com/your-repo/mental-health-assistant.git
cd mental-health-assistant
```

### **2. Install Dependencies**
```bash
pip install -r requirements.txt
```

### **3. Download NLTK Resources**
```python
import nltk
nltk.download('stopwords')
nltk.download('wordnet')
```

---
## Training the Model
### **Step 1: Preprocess and Train**
```python
python train.py
```
This script:
- Loads the dataset.
- Preprocesses text using **NLTK**.
- Extracts features using **TF-IDF**.
- Trains a **Random Forest Classifier**.
- Saves the trained model and vectorizer.

### **Step 2: Test Model with Example Queries**
```python
python test.py
```
---
## Deployment on AWS
This project can be deployed on **AWS Lambda & API Gateway** using Flask.

### **1. Create an AWS Lambda Function**
1. Package your code into a **ZIP file**:
   ```bash
   zip -r deployment.zip *
   ```
2. Upload the ZIP file to AWS Lambda.
3. Set the runtime to **Python 3.8+**.
4. Create a handler function (`lambda_function.py`):

   ```python
   from flask import Flask, request, jsonify
   import joblib

   app = Flask(__name__)
   model = joblib.load("mental_health_model.pkl")
   vectorizer = joblib.load("tfidf_vectorizer.pkl")
   label_encoder = joblib.load("label_encoder.pkl")

   def predict_emotion(query):
       vectorized_input = vectorizer.transform([query]).toarray()
       prediction = model.predict(vectorized_input)
       return label_encoder.inverse_transform(prediction)[0]

   @app.route("/predict", methods=["POST"])
   def predict():
       data = request.get_json()
       query = data.get("query", "")
       response = predict_emotion(query)
       return jsonify({"emotion": response})

   if __name__ == "__main__":
       app.run()
   ```
5. Deploy with **AWS API Gateway** to create an endpoint.

### **2. Access API**
Once deployed, you can access the API via:
```bash
curl -X POST "https://your-api-gateway-url.amazonaws.com/prod/predict" -H "Content-Type: application/json" -d '{"query": "I feel stressed and anxious."}'
```

---
## Future Improvements
- **Use Deep Learning (LSTMs, Transformers)** for better accuracy.
- **Integrate GPT-based models** for advanced response generation.
- **Expand dataset** to include more diverse mental health situations.

---
## License
MIT License. Free to use and modify.

---
## Contributors
- **Your Name** – [GitHub Profile](https://github.com/your-github)
- **Open for Contributions** 🚀