Chatbot architecture

RASA framework implementation research

definitions

• rasa is the framework. It is comprised of two distinct parts:
  • NLU which processes natural language understanding tasks to generate a language model,
  • and Core which processes the expression. both are trained separately.
• rasa x is the interface that helps users manage, train and deploy their rasa model.

setup

• get last chatbot_pptech repo update,
• from local scp repo to AWS instance at /home/ubuntu/chatbot/,
• from the aws_instance/chatbot/chatbot_pptech folder, install necessary requirements and train:

bash ./deploy.sh

run

• open 1st terminal on your computer and:

cd /home/ubuntu/chatbot/chatbot_pptech
screen
conda activate rasa
rasa run actions

• open 2nd terminal on your computer and:

cd /home/ubuntu/chatbot/chatbot_pptech
screen
conda activate rasa
rasa x

• you can close the terminals
• to reconnect to "screened" terminals open terminal and:
  • screen -ls lists terminals, get ids,
  • screen -r { id } returns to "screened" terminal

connect to chatbot

• password provided in terminal to connect
• go to <aws_instance/password>:port

documentations

• installation
• tutorial
• commands cheatsheet
• actions management
• community forum
• About Ubuntu Screen