Chatbot Challenge Platform

This repository contains code for the chatbot challenge platform. It contains these folders:

• chatbot-challenge-frontend The frontend of the chatbot challenge.
• escape-rooms The escape rooms which are Rasa agents. This folder contains the definitions of these dialog agents.

Preperation for Deployment:

• see deployment.md for more information about your preparations

Installation

Fixed Versions for bot:

• python 3.8.10
• rasa 3.4.2

For local Frontend-test:

• node-js (v20.2.0 ), npm ( 9.6.6)

Install Rasa:

Please refer to the official Rasa documentation for installation instructions.

https://rasa.com/docs/rasa/installation/installing-rasa-open-source/

Usually it makes sense to create a virtual environment for your python installation, such as Conda. In this way you can have multiple different Python environemnts in parallel that do not interfere with each other, in case you need Python for another project also.

Install Frontend:

First install nodejs and npm: https://nodejs.org/en/download/

git clone https://github.com/Chatbot-Challenge/Frontend.git
cd Frontend/chatbot-challenge-frontend/
npm install

Start it

Start a local Rasa server

We assume that you already trained a model. The -m parameter specifies the path to the trained model.

  cd <into-your-rasa-bot> 
  # or try out the example bot / that should work for you also
  cd Frontend/escape-rooms/example-bot

Rasa Server

• start rasa server
• explicitly set port to 5005
• make sure that api is available and reachable

rasa run \
    --enable-api  \
    --cors "*" \
    -p 5005

Rasa Actions

• in a second terminal
• start the actions server

rasa run actions

Test with curl

• check rasa server with curl
• on Linux run :

  curl -X POST -i http://0.0.0.0:5005/webhooks/rest/webhook -d '{"sender": "oha!", "message": "oha?"}'

• on Win$ run:

  curl -X POST -i http://localhost:5005/webhooks/rest/webhook -d  " { \"sender\":\"oha\" , \"message\": \"hi\" } "

• you should see something like :

  [{"recipient_id":"oha","text":"You can interact with: cell"}]

prepare frontend-config for local bot

• In Frontend/chatbot-challenge-frontend/config add a new room to the list of rooms :´
• make sure that you use the same port as specified above (rasa run -p 5005...)

config = {
    "name": "hello world",
    "rooms": [
        {
            "escape_room_name": "Lokal Demo",
            "chatbot_name": "local-bot",
            "user_name": "Oha Oha Oha",
            "messagebox_caption": "hmmm?...",
            "send_button_caption": "oha?",
            "user_name": "you",
            "id": "lokal",
            "background-image": "local-bot-background-image.jpg",
            "api_url": "http://localhost:5005",
            "welcome-message": "does it wörk?"
        },
      ...
      ]
    };

• Please note that configuring this file is only necessary for local testing.
  the deployment will use the config from your

Start the frontend

cd chatbot-challenge-frontend
npm run dev

Open http://localhost:5173/ in your web browser to see the user interface.

Configuration

chatbot-challenge-frontend/config.js contains the bot configuration. The configuration format is a JSON dictionary.

example:

{
    "name": "Under the sea",
    "id": "under-the-sea",
    "background-image": "under-the-sea-background-image.jpg",
    "api-url": "http://localhost:5005",
    "welcome-message": "You are under the sea."
}

Here is a list of available configuration parameters:

Parameter	Description	Required	Example Value
api_url	URL of the Rasa HTTP API	required	https://vm014.qu.tu-berlin.de/api5005
id	Unique ID of the escape room. Please derive this from your team name.	required	under-the-sea
escape_room_name	Name of the escape room	required	Under the sea
chatbot_name	Name of the chatbot that is displayed in the chat bubbles.	optional, default is "Game Master"	Neptun
user_name	Name of the user that is displayed in the chat bubbles	optional, default is "You"	Erika Musterfrau
messagebox_caption	Caption of the text input box in the chat window.	optional, default is "Type here what you want to do..."	Type here...
send_button_caption	Caption of the send button in the chat	optional, default="Do"	Swim
background-image	Name of the background image file in the folder public/background-images/. Please prefix with your id parameter to ensure that the filename is unique.	optional	under-the-sea.jpg
welcome-message	First message of the chatbot when the user starts the game	optional	You are under the sea.
chatbot-avatar	Image file to an avatar of the chatbot. Should be a rectangular image.	optional	jazz_club_avatar1.jpg
user-avatar	Image file to an avatar of the user. Should be a rectangular image.	optional	jazz_club_avatar2.jpg

Online installation

The chatbot challenge platform is installed live under https://vm014.qu.tu-berlin.de. The logfiles are accessible under https://vm014.qu.tu-berlin.de/logs . There is one log files for each escape room with the team number as filename.