TabulaRasa

interdisciplinary project HS Osnabrueck

Creating a rasa chatbot to answer FAQs and manage event registration for Kompetenzzentrum 4.0 Lingen

Setup Rasa X server (Docker)

1. Download the install script for Rasa X either for a specific version

curl -sSL -o install.sh https://storage.googleapis.com/rasa-x-releases/1.0.1/install.sh

or edge version

curl -sSL -o install.sh https://storage.googleapis.com/rasa-x-releases/latest/install.sh

2. Install Rasa X

sudo bash ./install.sh

3. Start up Rasa X in the Backround

cd /etc/rasa
sudo docker-compose up -d

4. Set up your password

cd /etc/rasa
sudo python3 rasa_x_commands.py create --update admin me <PASSWORD>

For more Information also see: (https://rasa.com/docs/rasa-x/installation-and-setup/install/docker-compose/)

Connect Rasa X with your Git Repository

1. Start Rasa X within your webbrowser and navigate to the "Connect to a reposetory" option

2. Clone your Git Repos URL with SSH and paste it in the "SSH for repository" field in Rasa X

3. For the filed "Target branch" choose "master"

4. Copy the SSH deply key from Rasa X and paste ist into your Git Repos Settings/Deploy keys. Title the key and give write accsess

5. Press "Verify connection" Button

Set up Docker Compose for the Docker Image with Spacy & Action Server

1. Download the Rasa X Docker-compose.yml (in this case for Version == 1.0.1)

wget -qO docker-compose.yml https://storage.googleapis.com/rasa-x-releases/1.0.1/docker-compose.ce.yml

2. Create a directory for the Rasa-Sevice Dockerfile (e. g. "rasa-service")

sudo mkdir rasa-service

3. Within the directrory create a Dockerfile

sudo nano Dockerfile

4. Paste this code to extend the Dockerimage with Spacy

ARG RASA_IMAGE
FROM ${RASA_IMAGE}

USER root

RUN pip install -U pip setuptools wheel
RUN pip install -U spacy

RUN python -m spacy download de_core_news_lg

USER 1001

5. Create a directory for the Rasa-Action Dockerfile (e. g. "rasa-action")

sudo mkdir rasa-action

6. Within the directrory create a Dockerfile

sudo nano Dockerfile

7. Paste this code to extend the Dockerimage for the Action Server

# Extend the offical Rasa SDK image
FROM rasa/rasa-sdk:2.8.2

# Use Subdirectory as working directory
WORKDIR /app

# Copy any additional custom requirements, if necesarry (uncomment next line)
# COPY actions/requirements.txt ./

# Change User
USER root

# Install extra reqirements for actions, if necesarry (uncoment next line)
# RUN pip install -r requirements.txt

# Copy actions folder to working directory
COPY ./actions /app/actions

# Change User
USER 1001

8. Within the rasa-action folder create a subfolder 'actions' for all custom actions files and the requirements.txt file. Paste the Pythone code into the files and the required libraries into the .txt file

cd etc/rasa/rasa-actions
sudo mkdir actions
sudo nano actions/actions.py
sudo nano actions/__init__.py
sudo nano actions/requirements.txt

9. Change permissions for rasa-service and rasa-action respectively

sudo chmod 775 -R rasa-service
sudo chmod 775 -R rasa-action

10. In the docker-compose.yml file, adapt the x-rasa-services: &default-rasa-service section

x-rasa-services: &default-rasa-service
  build:
    context: ./rasa-service
    args:
      RASA_IMAGE: "rasa/rasa:${RASA_VERSION}-full"
  restart: always
  volumes:
      - ./.config:/.config
.....

11. In the docker-compose.yml file, adapt the app section

  app:
    build:
      context: ./rasa-action
    restart: always
    expose:
      - "5055"
    depends_on:
      - rasa-production

12. Rebuild the Docker image

sudo docker-compose down
sudo docker-compose build
sudo docker-compose up -d

Rasa installation (local)

1. Create a virtual enviroment with python version 3.7

conda create --name tabalurasa python==3.7

2. Activate the virtual enviroment

conda activate tabularasa

3. Use an older version of pip due to backtracking runtime issue (https://stackoverflow.com/questions/65122957/resolving-new-pip-backtracking-runtime-issue)

python -m pip install --upgrade pip==20.2

4. Install rasa

pip install rasa

5. Copy this repository https://support.atlassian.com/bitbucket-cloud/docs/clone-a-git-repository/

6. Open the project in your favourite IDE

7. Create a settings.json in the project directory (root) with the following content:

{
    "python.pythonPath": "C:\\XXXXX\\XXXXX\\XXXXX\\XXXXX\\python.exe",
    "terminal.integrated.shell.windows":"C:\\Windows\\System32\\cmd.exe",
    "terminal.integrated.shellArgs.windows": ["/K", "C:\\XXXXX\\XXXXX\\XXXXX\\activate.bat C:\\XXXXX\\XXXXX\\XXXXX\\tabularasa"]
}

8. Adjust the path to the corresponding files e.g.:

{
    "python.pythonPath": "C:\\DEV\\miniconda3\\envs\\tabularasa\\python.exe",
    "terminal.integrated.shell.windows":"C:\\Windows\\System32\\cmd.exe",
    "terminal.integrated.shellArgs.windows": ["/K", "C:\\DEV\\miniconda3\\Scripts\\activate.bat C:\\DEV\\miniconda3\\envs\\tabularasa"]
}

9. Install spacy for advanced natural language processing

pip install spacy

10. Check your rasa version

rasa --version

This is my output:

This is my output:
Rasa Version      :         2.8.16
Minimum Compatible Version: 2.8.9
Rasa SDK Version  :         2.8.2
Rasa X Version    :         1.0.1
Python Version    :         3.7.11
Operating System  :         Windows-10-10.0.19041-SP0
Python Path       :         c:\dev\miniconda3\envs\tabularasa\python.exe

11. Train the rasa model

rasa train

12. Interact with the bot via terminal

rasa shell

For more information see the official rasa documentation: https://rasa.com/docs/rasa/

Access Rasa X server

1. Use the network of the hs osnabrück (e. g. via eduVPN)

2. Open 131.173.65.71/login in your favourite browser and login with credentials

UI maintenance

The UI is build with rasa webchat. A chat widget to deploy virtual assistants made with rasa or botfront on any website.

For more information see the rasa webchat project: https://github.com/botfront/rasa-webchat

The script which contains the body, the parameters and other features you will find here:

index.html

Access the ratings.csv

Option 1: From inside the container

1. Access Rasa X server (see above)
2. Open the container bash for navigation (replace the with the current id of the application container)

docker container exec -it <CONTAINERID> bash

3. View the content of the file by

cat ratings.csv

4. Close the bash with CTRL + D

Option 2: From outside the container

1. Access Rasa X server (see above)
2. Copy the file to your Local System (replace the with the current id of the application container)

sudo docker cp <CONTAINERID>:app/ratings.csv ~/ratings.csv

3. Navigate to copied file, where you will find the ratings

cd home/tabula

We recommend to access the ratings.csv from outside the container.