aws-lex-convo-bot-example

Reference implementation on building a conversational Amazon Lex Bot.

MoviePedia

Moviepedia ChatBot

Moviepedia Bot is a chat bot that helps you query information about a movie of your choice. We will be using AWS Lambda, which supports either of Node (0.10 or 4.3), Python (2.7), or Java (8) runtimes. This bot has been written in NodeJS and utilizes. TMDB for quering movies and returing desired information.

Steps to build the Moviepedia Bot

Step 1: Create the AWS Lambda function

Use this end point to upload the lambda code from s3: https://s3-us-west-2.amazonaws.com/re-invent-botworkshop/samples/moviePedia/moviePedia.zip

Instructions:

Zip the code that you downloaded. Note compress the files together not the folder. Now, go to AWS Lambda console. Create a new lambda function. Select "Blank Fuction" as a blueprint. in "Configure triggers" section press next. Now configure your lambda function.

Configure Function:

  1. Name your lambda function : moviePediaLogic
  2. Add Description - Lambda function for Moviepedia bot logic
  3. Runtime - Node.js4.3
  4. Code Entry - Upload the zip you downloaded
  5. Handler Section - Leave as default
  6. Select an existing role -  lambda_basic_execution
  4. Set time to 30 secs

5. Test your lambda function.

Configure the following test event to test your lambda function.

{
"messageVersion": "1.0",
"invocationSource": "FulfillmentCodeHook",
"userId": "user-1",
"sessionAttributes": {},
"bot": {
"name": "movieInfoApp",
"alias": "$LATEST",
"version": "$LATEST"
},
"outputDialogMode": "Text",
"currentIntent": {
"name": "movieInfo",
"slots": {
  "name": "Suicide Squad",
  "summary": "Director"
  },
"confirmationStatus": "None"
 }
}

The output should look like this:

{
 "sessionAttributes": {},
 "dialogAction": {
 "type": "Close",
 "fulfillmentState": "Fulfilled",
 "message": {
   "contentType": "PlainText",
   "content": "Director of Suicide Squad is/are: David Ayer"
   }
 }
}

Step 2: Creating your Bot

Create Amazon Lex IAM Role: lex-exec-role

Go to Identity and Access Management (IAM) console. In role name, use a name that is unique within your AWS account (for example, lex-exec-role).

In Select Role Type, choose AWS Service Roles, and then choose AWS Lambda.

Note In the current implementation, Amazon Lex service role is not available. Therefore, you first create a role using the AWS Lambda as the AWS service role. After you create the role, you update the trust policy and specify Amazon Lex as the service principal to assume the role. In Attach Policy, choose Next Step (that is, you create a role without any permissions).

Choose the role you created and update policies as follows:

In the Permissions tab, choose Inline Policies, and then attach the following custom policy.

{ 
"Version": "2012-10-17", 
"Statement": [ 
{ 
  "Action": [ 
    "lambda:InvokeFunction"
  ], 
  "Effect": "Allow", 
  "Resource": "*" 
  } 
 ] 
}

In the Trust Relationships tab, choose Edit Trust Relationship, and specify the Amazon Lex service principal ("lex.amazonaws.com"). The updated policy should look as shown:

{
 "Version": "2012-10-17",
 "Statement": [
{
  "Effect": "Allow",
  "Principal": {
    "Service": "lex.amazonaws.com"
  },
  "Action": "sts:AssumeRole"
  }
 ]
}

1. Create Amazon Lex Bot

Go to Amazon Lex console on create your Amazon lex bot page. Select custom app and provide the following information, then choose Create.

Remember to use a unique name for your bot

Provide The Following Information:

  1. Bot Name: moviePediaInfo
  2. Choose an output voice - Salli
  3. Set Session Timeout - 5 mins
  4. Add AMazon lex basic role to your Bot app - lex-exec-role

Step 3: Creating your Bot Conversations

2. Create Slots

In this section you create the two slots types for our ChatBot.

Remember to use a unique name for your slots

Create your First slot

Our first slot type is for the user to input a movie name he wants to inquire on.

 Slot Type Name : moviePediaInfomovieName 

Configure your first slot

Click on the slot type you created in the previous step and then add description to your slot type. Let's add in some sample movie names as values here.

Save your first slot

Now that you have configured your Slot save your Slot configuration.

Create your second slot

Our second slot type is for the user to input what information is he looking for. In this example, our lambda function where the bot logic sits is quering omdb to return values for ratings, votes, actors, plot, director, year of release.

 Slot Type Name : moviePediaInfomovieDetails 

Remember to use a unique name for your slots.

Configure your second slot

Click on the slot type you created in the previous step and then add description to your slot type. Let's add in values for our queries here.

Save your Second slot

Now that you have configured your Slot save your Slot configuration.

3. Create Intent

Intents help us map our conversations to our slots. Our MoviePedia Chatbox requires creation of a single intent.

4. Configure Intent

Configure some sample utterances for you to start interacting with the bot.

Sample Utterances:

  1. Hi
  2. Hello Moviepedia
  3. Tell me about a movie
  4. Tell me about a {name}

Map Slots

Add the following name, slot type and prompt for our ChatBot. Note the sequence. Enter the values for the slot type corresponding to moviename first.

Enter:

Name Slot Type Prompt

---

name moviePediaInfomovieName Welcome to MoviePedia! Confirm the movie name you want to know about?

Name Slot Type Prompt

---

summary moviePediaInfomovieDetails What information are you looking for? Year, Plot, All?

Remember check the checkbox as both are slots are required for our ChatBot {#toc_30}

Add Lambda to the Intent

We need to link our Chatbot to the lambda function. Type in an appropiate goodbye message for your users.

Save The Intent

Now that you have configured your Intent scroll up and save your Intent configuration.

Error Handling

We will be customizing the error message for our bot users. Click on error handling. You will notice there are default values set there. Delete the default values and add in a customized message like. Add this to the Prompts section:

Welcome to Moviepedia. Type "Tell me about a movie"

Step 4. Build

Once you have configured your moviePedia chatbot. Click on build to build your chatbot.

Step 5. Test App

Your browser does not support the video tag.

As, the build suceeds it's time for you to test the chatbot.

Step 6. Publish App

Once your Bot is build and tested the next step is to publish our Bot and make it avaialble on Slack, Facebook etc. In this example we will be deploying our Bot on Facebook. To publish your Bot you need to create an Alias. In this example we create an alias for "prod".