Switch branches/tags
Nothing to show
Find file History
Latest commit 74fd8cd Nov 1, 2018
Permalink
..
Failed to load latest commit information.
hr_sjefen.zip Add files via upload Oct 30, 2018
readme.md Update readme.md Nov 1, 2018

readme.md

Make Your Own Chatbot!

This short tutorial will guide you through making your very own text based game inside a chat bot. Google's Dialogflow is the chat bot technology of choice in this tutorial.

Prerequisites:

  • Google Account

Task

Add intents so that the chatbot user can exit the room.

Setup

  1. Go to Dialogflow.com
  2. Click sign up for free
  3. Log into your Google Account
  4. You should now see the welcome page to Dialogflow.
    • Click create agent
  5. You should now see agent setup page
    • Give the agent a name
    • Use English as default langauge. (Note: Norwegian is also a selection, but since we have used English in the premade data this should be used)
    • Click create

Guided tour around Dialogflow

  1. Assumption: you have a new and clean agent.
Import prebuilt agent
  1. Click prebuild agents
  2. Import the coffe shop agent
Creating new intent

An intent is a collection of phrases from the users mapped to responses.

  1. Click the intents menu point
  2. Click create intent
  3. Add a the name "definition of affogato"

A training phrase is an example of input that the user might want to write to hit this intent.

  1. Add training phrases
    • what is the definition of affogato?
    • what is the definition of an affogato?
    • what is an affogato?
    • what is affogato?
    • affogato

A response is the answer that the chat bot gives the user if he/she hits this intent.

  1. Scroll down and add responses
    • An affogato (Italian for "drowned") is an Italian coffee-based dessert. It usually takes the form of a scoop of vanilla gelato or ice cream topped or "drowned" with a shot of hot espresso.

Context is used to remember things from earlier in the dialog.

  1. Scroll back up and add output context
    • text should be definition_of_coffee
    • click the number and set the timeout to 1
  2. click save
    • when you see the button turn green the intent should be saved and the test chat is ready.
Testing the new intent
  1. To the far right of the site there is a try it now field.
    • click field and type in "affogato"
    • the output should then be the response added to the intent and a output context with a timeout of 1.
Adding a follow up intent
  1. To follow up the previous affogato intent we can now create a new intent

    • click the + button next to the intents menu point.
  2. Add the name "definition of karsk"

  3. Add input context

    • text should be "definition_of_coffee"
    • there should also be an output context with the same text.
  4. Add training phrases

    • how about karsk?
  5. Add response

    • "Karsk is a Norwegian cocktail from the Trøndelag region containing coffee together with moonshine and sometimes a spoon of sugar."
  6. click save and test it in the chat.

Adding a entities

An entity is used to identify and pull out information from a phrase.

  1. Click the entities menu point.
  2. Click the @drink entity
  3. Add a new row to the bottom
    • Name: Lemonade
    • Values: Lemonade, Lemon juice, citrus juice
  4. save and try it out the chat by typing
    • can i have a citrus juice.

Tasks

Using pre built agents from Google

Collecting contact information

Purpose: Hand out t-shirts after collecting contact information.

  1. Create an intent that starts the dialog
    • add training data with questions ex. "i want a free t-shirt!"
  2. Create an intent that collects contact information after the previous intent has triggered the dialog.
  3. create parameters for first and last name, e-mail and phone number (use system defined entities). All parameters are required.
  4. Add norwegian names and email in the phrases. Connect them to entities if they are not defined. Norwegian first names can be connected to @sys.given-name
  5. After the information have been collected, thank the user and say the t-shirt is sent.

Support

Setup: Go to the prebuilt agent section and find the agent called support and click import

  1. Let the user report a bug
  2. Ask if the bug is serious
  3. ask if the user has a work around.
  4. fill out the answers on the other intents.

Flights

Setup: Go to the prebuilt agent section and find the agent called flights and click import Task: Expand the features of this chatbot by:

  1. adding a way to order parking
    • It should be possible to order parking inside or outside
    • if its inside there should be a way to order a car wash as well.
  2. order a rental car
    • how many days?
    • what type of car? (large, medium, small, sport)
    • what color?
    • with or without childrens seat.

Jokes

Setup: Go to the prebuilt agent section and find the agent called jokes and click import Purpose: Create knock knock jokes!

  1. Create a knock knock joke where the bot asks knock knock
  2. Flip it around! Create a dialog where the user asks, knock knock and the bot joins in answering whos there... etc. (hint: can this be done with 1 intent?)
  3. Expand task 1. by adding a variation of knock knock jokes.

Tourism

Setup: Go to the prebuilt agent section and find the agent called Tourism and click import

  1. Add answers on the intents that are added
  2. Expand the bot to handle Norwegian places

FAQ

Setup: Go to the prebuilt agent section and find the agent called FAQ and click import

  1. create your own FAQ bot!
  2. You can use the prebuild agent for referance