Conversational Agents

This repository contains the group project for the TU Delft DSAIT4065 (previously CS4270) Conversational Agents course. The project implements a psychological doctor agent system powered by conversational AI technologies.

Table of Contents

• Conversational Agents
  • Table of Contents
  • 1. Prerequisites
  • 2. System Interface
  • 3. Setup Instructions
    • 3.1 Environment Variables Setup
    • 3.2 Facial Emotion Detection Setup
      • 3.2.1 Option 1: Using venv
      • 3.2.2 Option 2: Using Conda
    • 3.3 Docker Deployment
  • 4. Administration
    • 4.1 MongoDB Management
  • 5. Troubleshooting
    • 5.1 Docker Commands

1. Prerequisites

• Docker and Docker Compose
• Python 3.10 or higher (for local deployment)
• OpenAI API key (for conversation capabilities)

2. System Interface

The psychological doctor agent features a Streamlit-based web interface that provides the following functionality:

• User registration and login system
• Text-based conversation with the agent
• Audio input capability via microphone
• Real-time facial emotion detection (Not shown in the demo figure below)
• Text-to-speech response playback

Below is a screenshot of the user interface:

3. Setup Instructions

3.1 Environment Variables Setup

Create a .env file in the ./conversation directory based on the provided ./conversation/.env.default template.

3.2 Facial Emotion Detection Setup

Note: Since camera access is challenging within Docker containers, this service runs locally with Docker mapping localhost to internal container hosts through extra_hosts.

To set up the facial emotion detection component, follow one of the options below:

3.2.1 Option 1: Using venv

1. Create a Python virtual environment:

   python -m venv venv

2. Activate the virtual environment:

   • For Windows:

     venv\Scripts\activate

   • For macOS/Linux:

     source venv/bin/activate

3. Install the required dependencies:

   pip install -r deepface/requirements.txt

4. Run the facial emotion detection service:

   python deepface/app.py

3.2.2 Option 2: Using Conda

1. Create a Conda environment with Python 3.10:

   conda create -n emotion_detection python=3.10
   conda activate emotion_detection

2. Install the required dependencies:

   pip install -r deepface/requirements.txt

3. Run the facial emotion detection service:

   python deepface/app.py

3.3 Docker Deployment

1. Start all services (MongoDB and conversation) using Docker Compose:

   docker-compose up -d

   To rebuild images when changes are made:

   docker-compose up -d --build

2. Once deployed, access the Streamlit conversation interface at: http://localhost:8501

4. Administration

4.1 MongoDB Management

1. Connect to the MongoDB database:

   docker exec -it mongo mongosh -u admin -p password --authenticationDatabase admin

2. View conversation data:

   use conversations
   db.users.find().pretty()

5. Troubleshooting

5.1 Docker Commands

Common Docker commands for troubleshooting:

• Access container shell:

  docker exec -it <container_id_or_name> /bin/bash

• View container logs:

  docker logs <container_id_or_name>

• Restart services:

  docker-compose restart

• View logs for all services:

  docker-compose logs -f

• View logs for a specific service:

  docker-compose logs -f <service_name>