Introduction to Forms

In this segment we will teach you how to make custom forms. Rasa Forms are a way to create collect slot values from the user in a structured way.

Video#

Code#

There are some important changes that we've made to our domain.yml file. You can see them highlighted in the code block below.

version: '2.0'
session_config:
  session_expiration_time: 60
  carry_over_slots_to_new_session: true

intents:
  - greet
  - deny
  - request_names
  - goodbye
  - affirm
  - mood_great
  - mood_unhappy
  - bot_challenge

forms:
  name_form:
    first_name:
    - type: from_text
    last_name:
    - type: from_text

slots:
  first_name:
    type: text
    influence_conversation: true
  last_name:
    type: text
    influence_conversation: true

responses:
  utter_greet:
  - text: Hey! How are you?
  utter_cheer_up:
  - text: 'Here is something to cheer you up:'
    image: https://i.imgur.com/nGF1K8f.jpg
  utter_did_that_help:
  - text: Did that help you?
  utter_happy:
  - text: Great, carry on!
  utter_goodbye:
  - text: Bye
  utter_iamabot:
  - text: I am a bot, powered by Rasa.
  utter_ask_first_name:
  - text: What is your first name?
  utter_ask_last_name:
  - text: What is your last name?
  utter_submit:
  - text: Ok. Thanks!
  utter_slots_values:
  - text: I will remember that your name is {first_name} {last_name}!

Form Definition#

First, the domain file has a form defined. It's defined in this part:

forms:
  name_form:
    first_name:
    - type: from_text
    last_name:
    - type: from_text

We're declaring that our form has two slots that we'd like to fill; first_name and last_name.

Slot Definition#

Second, the domain file has slots defined. Note that the names of these slots correspond with the names in our form definition.

slots:
  first_name:
    type: text
    influence_conversation: true
  last_name:
    type: text
    influence_conversation: true

Note that these two slots have types attached. In this case we're using strings but Rasa allows for many different slot types.

Utterance Definition#

Finally, we also add some utterances in the domain.yml file. These are used in our stories.

  utter_ask_first_name:
  - text: What is your first name?
  utter_ask_last_name:
  - text: What is your last name?
  utter_submit:
  - text: Ok. Thanks!
  utter_slots_values:
  - text: I will remember that your name is {first_name} {last_name}!

You may have also noticed that some utterances follow a specific naming convention utter_ask_<slotname>. These utterances will be picked up by our form to ask for a slot value. This is explored in more detail in the next video.

The utter_slots_values utterance uses slot values to generate a response. This allows us to confirm what the user has provided as names.

Stories#

The dialogue in the video is recorded via rasa interactive and it will produce a story in our stories.yml file. Although it's not shown in the video, we wanted to share the story here.

- story: interactive_story_1
  steps:
  - intent: greet
  - action: utter_greet
  - intent: request_names
  - action: name_form
  - active_loop: name_form
  - slot_was_set:
    - requested_slot: first_name
  - slot_was_set:
    - first_name: vincent
  - slot_was_set:
    - requested_slot: last_name
  - slot_was_set:
    - last_name: vincent-mcvincent
  - slot_was_set:
    - requested_slot: null
  - active_loop: null
  - action: utter_submit
  - action: utter_slots_values

Links#

• GitHub repo
• Rasa Open Source Docs
• Conversational AI with Rasa: Forms Overview
• Conversational AI with Rasa: Custom Forms

Exercises#

Try to answer the following questions to test your knowledge.

1. Can you use slots in responses defined in the domain file?
2. How can you indicate that a slot was set in the stories.yml file?
3. When slots appear in a stories.yml file, do they influence the next action prediction? If so, do they do it all the time?