Skip to content

Latest commit

 

History

History
196 lines (139 loc) · 4.81 KB

README.md

File metadata and controls

196 lines (139 loc) · 4.81 KB
Raw

Demo AWS Lex Chatbot

Introduction

Amazon Lex is a service for building conversational interfaces into any application using voice and text.

It can be embedded in any website using a js snippet or server as a whole application

The project is intergrated with some of AWS toolkits in order to build a dynamic bot with dynamic data without rebuilding the bot

Services Used

  • Lex (Chatbot)
  • Lambda ( Handle further Initialization, Validation and Fulfillent )
  • SNS ( Send SMS )
  • Cloud Formation (Host Lex bot)
  • S3 (Create bukets)
  • DynamoDB (Query and Put data)
  • IAM (Config roles and permisisons)
  • CloudWatch (Log errors)
  • Cognito (create pool)

Trigger Intents

My Booking Movie Ticket Boot has 3 intents as listed below

BookTickets

Utteraces

Slots

GetTheater

Utteraces

Slots

Help

Buit-in Lex intent

Lambda Functionex (Chatbot)

  • Lambda ( Handle Initialization, Validation and Fulfillent further)
  • SNS ( Send SMS )
  • Cloud Formation (Host Lex bot)
  • S3 (Create bukets)
  • DynamoDB (Query and Put data)
  • IAM (Config roles and permisisons)
  • CloudWatch (Log errors)
  • Cognito (create pool)
  • Connect (Phone Calls)

Trigger Intents

My Booking Movie Ticket Boot has 3 intents as listed below

BookTickets

Utteraces

Slots

GetTheater

Utteraces

Slots

Help

Buit-in Lex intent

Lambda Function

Perform further initialization, validation and fulfillment in lex.

If the validation process sends a success message, the chatbot will ask the next question to fulfill the remaining slots. Otherwise, it will ask for confirmation then end the session.

If any slot has data that doesn't exist in the database, it will send an Elicit Slot Action Type and ask the user to input data again.

Max invalid data entering per intent is 2 (customizable). If entered data is invalid in 3 rounds of input, end the session.

Validation

It performs the following user inputs validation.

  • MovieName slot should be in a pre-defined list ( in the real application it will call a backend server and get an available list of movies)

  • TheaterName slot should be in DynamoDB associated with MovieName slot (movie table)

  • Users can book up to 1-month tickets in advance from today's sales

  • The number of ticket bookings is bound from 1 to maximum booking up to 10

  • Implemented for basics validation for mobile number

  • When the bot has all the checked slots, check if movie_id has existed in the database. If it doesn't, end the current session

Confirmation state

Yes

  • Save the order to order table
  • If the number starts with +1 then send a confirmation SMS with booking details
  • Push a confirmation prompt

No

  • Push a confirmation prompt

Lambda Roles

{
    "Version": "2012-10-17",
    "Statement": [
        {
            "Sid": "VisualEditor0",
            "Effect": "Allow",
            "Action": [
                "logs:CreateLogStream",
                "dynamodb:PutItem",
                "logs:CreateLogGroup",
                "logs:PutLogEvents"
            ],
            "Resource": [
                "arn:aws:logs:*:*:*",
                "arn:aws:dynamodb:*:*:table/DummyOrder"
            ]
        },
        {
            "Sid": "VisualEditor1",
            "Effect": "Allow",
            "Action": "dynamodb:Query",
            "Resource": "arn:aws:dynamodb:*:*:table/DummyMovie*"
        },
        {
            "Sid": "VisualEditor2",
            "Effect": "Allow",
            "Action": "sns:*",
            "Resource": "*"
        }
    ]
}

Database Structure

DummyMovie Table

Indexes

  • movieName (String) (Partition Key)
  • theaterName (String) (Sort Key)

DummyMovie Order

Indexes

  • userId (String) (Partition Key)

Cloud Formation for LEX

The UI of the chatbot is created by using the aws-lex-web-ui project. The project provides a cloud formation stack that creates a static website on the S3 bucket for the chatbot web UI.

Demo