# Text-Based Chatbot Performance Analysis

This project implements an end-to-end performance analysis system for text-based chatbots using Machine Learning. It features a Bidirectional LSTM model with an Attention mechanism to evaluate chatbot responses based on provided context and facts.

## Features
- **Exploratory Data Analysis**: Visual insights into engine distribution and performance correlations.
- **Advanced LSTM Model**: Uses Bidirectional LSTM and Attention layers for high-accuracy evaluation.
- **Context-Aware Prediction**: Evaluates responses not just on linguistics but also on factual consistency.
- **Modern Web Interface**: Glassmorphic UI with real-time performance analytics.
- **Flask Backend**: Robust API for model inference.

## Project Structure
- `train_model.py`: Training pipeline for the advanced model.
- `app.py`: Flask server for real-time predictions.
- `explore_data.py`: EDA script for dataset visualization.
- `BP_MHS_V1.csv`: The core dataset.
- `templates/` & `static/`: Frontend assets.

## How to Run
1. Install dependencies:
   ```bash
   pip install -r requirements.txt
   ```
2. Train the model:
   ```bash
   python train_model.py
   ```
3. Start the application:
   ```bash
   python app.py
   ```
4. Access the UI at `http://127.0.0.1:5000`.

## Model Insights
The system uses an Attention mechanism to focus on critical parts of the facts and responses, ensuring the expert verdict is both accurate and contextually relevant.