RASA: An Overview 

What is RASA?

• An open source framework used to build customized chatbots

Components of RASA:

• Rasa NLU

• Rasa Core

RASA NLU:

• It helps the chatbot to interpret and understand user question

Example: User's Question, ‘Are you a human?’

RASA NLU helps bot to understand the question

Rasa Core:

• It decides the action to be performed by bot.

• Also, it makes the bot to learn continuously and improves the prediction response for next intent.

Example: RASA core helps bot to answer user's question

'No, I am not a human. I am Bot made with RASA'

Why RASA?

• Easy to customize and flexible

• Interactive learning

How to install RASA:

https://rasa.com/docs/rasa/installation/environment-set-up/#!

Nomenclatures in RASA:

1. Intent:

• It gives the intention of the input provided by user

Example:

a. User's Input: ‘Hi’ or ‘Hello’ or ‘Good morning’

Here ‘Intent’ is ‘Greet’ ------> User is greeting the bot

b. User's Input: ‘Yes’ or ‘Correct’ or ‘Y’ or ‘Accept’

Here 'Intent’ is ‘Accept’ --------> (User is providing his acceptance)

2. Entities:

• It is a piece of information that can be extracted from a user's input.

Example:

a. User's Input: ‘I want to fly from Chennai to Delhi’

Here the ‘Entity’ is ‘City’ --------> Chennai & Delhi

b. User's Input: ‘Show me an Indian restaurant’

Here the ‘Entity’ is ‘Cuisine’ --------> Indian

3. Slots:

• It is a memory of a bot which is used to store input provided by user

Example:

To play a game 'Rock, Paper & Scissors', user choice's will be saved.

slots:
  choice:
    type: categorical
    values:
      - rock
      - paper
      - scissors

In above code, 'choice' is a 'slot' in which one of the values will be stored.

4. Actions & Responses:

• It is a response from bot for an intent

Example:

a. For intent ‘Greet’, bot's response is ‘Hey’, ‘Hello’, ‘Hi’ or ‘Good Morning’

b. For intent ‘Goodbye’, bot's response is ‘Have a nice day’, ‘See you later’ or ‘Bye’

5. Stories:

• These are sample interaction between the user and bot.

• Usually, stories are defined in terms of intent captured and actions performed.

Example:

Story: play game

Steps:

Intent: Greet                                                       ---- > User's Intent

Action: Greet   ---- > Bot's Response

Action: Ask ‘Do you want to play’  ---- > Bot's Response

Intent: Get the input from user ‘y’ or ‘n’      ---- > User's Intent

Action: Play game                                              ---- > Bot's Response

6. Domain:

• It defines all the intents, entities, slots, responses and actions that bot should know.

Steps to create a bot:

1. Initialize the project

2. Define Stories

3. Define Intents

4. Define Actions

5. Define Domain

6. Train Model

Want to create a new bot which plays Rock Paper & Scissor with you? See below video

Conclusion:

This blog gives an definitions of basic terminologies in RASA. However, this is not an end. So, Happy Learning!!!!