Skip to content

Latest commit

 

History

History
92 lines (66 loc) · 4.44 KB

configure_chatbot.md

File metadata and controls

92 lines (66 loc) · 4.44 KB
Raw

Building a Chatbot using Lex

Lets create a Halloween chatbot which is capable of having a simple conversation. The goal of the chatbot is to collect a user's name, the costume they are wearing and know if the user would like to have a chocolate. Before continuing any further make sure that you are familiar with AWS Lex vocabulary. For details on AWS Lex visit this documentation page

First lets define a blueprint for our chatbot, that is, the Intents, Slots, Utterances, Prompts etc,. for our chatbot.

Blueprint:

    Intent : GetInfo
    Slots : 
        - Name (AMAZON.US_FIRST_NAME built-in type) 
        - Costume (AMAZON.FictionalCharacter built-in type) 
    Utterance :
        - "Hello"
        - "Hi"
        - "Hello, I am {Name}"
        - "Hi, I am {Name}"
        - "Hi, Who are you " 
        - "Hi, What are you "
        - "Hi, What is you name "
        - "Hello, Who are you " 
        - "Hello, What are you "
        - "Hello, What is you name "
    Prompts :
        Name prompts :
            - "Hi! I'm Candy Monster. What is your name ?"
        Costume prompts :
            - "Who are you dressed up as {Name}?"
        Confirmation prompt :
            - "I think {Name} and {Costume} both deserve a chocolate. Do you want a chocolate?"
        Cancel (if the user says "no") prompt :
            - "Okay. That's probably better for you anyway. Good day!"
    Fulfillment message:
        Goodbye message:
            - "Enjoy you chocolate!"

Steps to create the chatbot:

Before working on the chatbot, lets build a lambda which we would use while building the chatbot. Login to the aws console with your credentials and visit this page and click on Create function, then click on Author from scratch. This should take you to a page which look like the image below. Enter the name for the lambda as puppetbot-lambda and for role select Create a custom role. This should take you to a new page, just click Allow. Then you are be back on the previous page, just click on create function.

Now select the Runtime as Python 2.7 and change the name of Handler from lambda_function.lambda_handler to main.handler. Copy and paste the following code into the Function code. Hit Save

def handler(event, context):
    return {
                "sessionAttributes": event['sessionAttributes'],
                "dialogAction": {
                    "type": "Close",
                    "fulfillmentState": "Fulfilled",
                    "message": {
                        "contentType": "PlainText",
                        "content": "Enjoy"
                    }
                }   
            }

Now lets build our chatbot. Visit this page, it should look something like this.

Click on Create button and then select Custom bot. Lets name our bot as puppetbot and choose Output voice as Ivy. I personally like Ivy voice for this project as it has a robotic flavor to it. Our chatbot's conversion is small and shouldn't exceed more than 3 mins, so lets set Session timeout as 3 mins. For COPPA select No for now. Click on Create button, this create our chatbot.

Once the chatbot is created select Create intent and then select Create new intent. As we want to create an intent with name GetInfo enter it into the input box and click on Add button. Add all the Utterances from our blueprint into the Sample utterances section.

Now fill the Slots and Prompts from our blueprint into the respective sections.

Now click on Fulfillment section, select AWS Lambda function, select puppetbot-lambda which we created earlier and enter "Enjoy you chocolate!" as Goodbye message. Hit Save Intent button. We successfully created the chatbot, now lets publish it.

Click on Build. Once the build is finished click on Publish button, enter dev in Create as alias input and hit Publish. Now our chatbot is up and ready!