KI-Campus / chatbot-kim

Source code for the AI Campus Chatbot
https://ki-campus.org
MIT License
4 stars 1 forks source link
chatbot

KI-Campus Chatbot "KIM"

Training of the Model

Choose the folder with the chatbot in English (KI-Campus_en) or in German (KI-Campus_de)

    rasa train

Configuration

Course Recommender Endpoint

NOTE Currently only supported for German chabot (rasa/KI-Campus_de/)

Create (or modify) the configuration file kic_recommender.yml in directory

    rasa/KI-Campus_de/

and add/modify configuration entry for Course Recommender Endpoint with the base URL for the endpoint and the access token:

# This file contains the custom service endpoints your bot can use.

# recommender service (DFKI) configuration
recommender_api:
  url: "<base URL for recommender service endpoint>"
  token: "<recommender access token>"

Usage

Start the Rasa Server

    rasa run --enable-api

Start the Action Server

    rasa run actions

NOTE the actions must not be started from within the actions/ sub-directory, but the rasa project's root directory (e.g. rasa/KI-Campus_de/), otherwise not all actions may be automatically started.

Start for Development

Change configuration to 'for local development' endpoint in endpoints.yml in sub-directories

    rasa/KI-Campus_de/
    rasa/KI-Campus_en/

then (within the respective directory) start the chatbot shell with

    rasa shell

for testing logged-in user, get your access-token from the ki-campus website and start shell with

    rasa shell --conversation-id <access-token>

Docker

In the outer project structure run:

Docker Compose (de)

    docker-compose -f docker-compose_de.yml -p kicampus_de up --build

Docker Compose (en)

    docker-compose -f docker-compose_en.yml -p kicampus_en up --build