🤖 Friendly Chitchat Chatbot with RASA & Angular

This repository contains the code of the Friendly Chitchat Chatboot, implemented with RASA & Angular. You can ask chitchat wit your chatbot about a lot of things !

Usage

Pre-requisites

Installing rasa and Angular :

pip install rasa

Angular :

npm install -g @angular/cli

Backend Chatbot

In order to start the chatbot endpoint, run the following command inside the chatbot folder :

rasa run --model models --enable-api --cors "*" --debug -p 5005

Or run the following two commands to start the backend in a shell script :

sudo chmod +x chatbot.sh
source chatbot.sh

Frontend App

After running the backend, we can start the Angular frontend app with the following command inside the frontend folder :

ng serve

Or with the following two commands to run it in a shell script :

sudo chmod +x frontend.sh
source frontend.sh

Chatbot data

The data that has been used to train is contained in the nlu.yml & stories.yml. It contains 87 intents (greetings, user.bored, appraisal.thank_you, ...) written in english. The pipeline it goes through is the follow :

pipeline:
  - name: WhitespaceTokenizer
  - name: RegexFeaturizer
  - name: LexicalSyntacticFeaturizer
  - name: CountVectorsFeaturizer
  - name: CountVectorsFeaturizer
    analyzer: "char_wb"
    min_ngram: 1
    max_ngram: 4

Chatbot Classifier

The chatbot uses the DIETClassifier and the ResponseSelector in order to selec the right answers. The full configuration used for training is the following :

language: en
pipeline:
  - name: WhitespaceTokenizer
  - name: RegexFeaturizer
  - name: LexicalSyntacticFeaturizer
  - name: CountVectorsFeaturizer
  - name: CountVectorsFeaturizer
    analyzer: "char_wb"
    min_ngram: 1
    max_ngram: 4
  - name: DIETClassifier
    epochs: 100
  - name: EntitySynonymMapper
  - name: ResponseSelector
    epochs: 100

policies:
  - name: MemoizationPolicy
  - name: TEDPolicy
    max_history: 5
    epochs: 100