Chatbot Development Process

Requirements

These are abstract notes I brainstormed first.

To set up your environment for building chatbot tools, make sure you have the following components installed:

1. Docker
2. Miniconda
3. JupyterLab

Miniconda setup

(flow: activate +jupyter lab port)

mkdir -p ~/miniconda3
wget https://repo.anaconda.com/miniconda/Miniconda3-latest-Linux-x86_64.sh -O ~/miniconda3/miniconda.sh
bash ~/miniconda3/miniconda.sh -b -u -p ~/miniconda3
rm -rf ~/miniconda3/miniconda.sh

~/miniconda3/bin/conda init bash
~/miniconda3/bin/conda init zsh

source ~/.zshrc

Create a Python environment for chatbot:

conda create -n chatbot python=3.7

Activate the chatbot environment and verify Python:

conda activate chatbot
python --version

JupyterLab Setup

Using miniconda, get JupyterLab:

conda update -n base -c defaults conda
conda clean -tipsy
conda install -c conda-forge jupyterlab

conda install jupyterlab 

pip install jupyterlab

Create your workspace and navigate to it:

mkdir chatbot && cd kouloobot

Start a new lab using:

jupyter lab

You can also access using the following:

jupyter lab --ip 0.0.0.0 --port 5050

sudo ufw allow 5050

Screen Package Use Case

You use some server shell and want it to play in the background. Then get screened.

1. Install the "screen" package using apt if you haven't already:

   sudo apt install screen

2. To start a new screen session named "jupyterlab", open a terminal and type the following command:

   screen -S jupyterlab

   This will take you to a new screen session where you can run other commands or shells.

3. To detach from the current screen session and return to the main terminal:

   Press Ctrl + A and then Ctrl + D.

4. If you've detached from the screen session and want to get back to it, use the following command:

   screen -r jupyterlab

Building Chatbot Tools

You can build chatbot tools using the following technologies:

1. Dialogflow
2. Botpress (Free to use, Docker-based)

Setting up Jupyter Lab

1. Open Jupyter Lab.
2. Go to your terminal in the chatbot directory.
3. Install the 'wget' and 'zipurl' packages.
4. Install 'zip' and 'unzip' packages using the command: sudo apt install zip unzip.

Chatbot Configuration

In the chatbot configuration, consider the following steps:

1. Care about Docker Compose files, and select the one for the community version.
2. Ensure you update the EXTERNAL_URL in your Compose file. If you are using a virtual machine, change the IP accordingly.
3. Update the password for PostgreSQL (psql) if required.

Docker Volumes

For Docker Compose, you need to create and configure directories:

1. Create directories for Botpress data and language using mkdir -p ./botpress/data and ./botpress/language.
2. Give permission to both directories using sudo chmod -R 777 for both directories.
3. Update the volume paths in Docker Compose to botpress_data and botpress_language. This ensures the volumes are attached to Docker, not stored locally.

Dockerfile Update

Inside the Botpress directory, update the image version from DockerHub in your Dockerfile.

Testing the Environment

Refer to the README of your Docker Compose configuration for the exact command to run the community version. Run it using sudo.

docker-compose -f docker-compose-community.yaml up --build

Keeping Botpress Maintained

To keep Botpress up to date:

1. Update the version in your Dockerfile FROM.
2. Add the flag --auto-migrate to the CMD to enable automatic migrations.

Sometimes, a fresh installation is required. To do this:

1. Clean your resources by turning down all containers using sudo docker-compose down.
2. Remove any containers using sudo docker rm -f $(sudo docker ps -a -q).
3. Terminate all resources with sudo docker system prune -a.

Now you are ready to reinstall Botpress.

Designing Chatbot Flow

When designing your chatbot's flow, consider the following:

• Implement proper delay times using "execute code" and run utility functions.
• Organize the "on enter" order effectively.
• Store data from conversations in storage functions.
• Provide file attachments for the user.

AmazonLinux

Prerequisites

sudo yum install -y docker
sudo systemctl enable docker
sudo yum install -y git

ssh -i ~/rsa.pem ec2-user@ip

Screen Package Use Case

• Install the "screen" package using yum if you haven't already:

  sudo yum install screen

• To start a new screen session named "jupyterlab," you can give it a name open a terminal and type the following command:

  screen -S jupyterlab

• To detach from the current screen session and return to the main terminal:
  • Press Ctrl + A and then Ctrl + D.
• If you've detached from the screen session and want to get back to it, use the following command:

  screen -r jupyterlab

Volume config

mkdir -p ./botpress/data
mkdir -p ./botpress/language

sudo chmod -R 777 ./botpress/data
sudo chmod -R 777 ./botpress/language

In volumes, add the following

volumes:
  - ./botpress/data:/botpress/data

volumes:
  - ./botpress/language:/botpress/lang

Workflow Representation

It's essential to apply your scenario on paper and later translate it into a chatbot flow using nodes connected to one another.