Making AI BOT service easy, by AWS
Lets build a Conversational interface between a User[Human] and a AI powered Bot system.
AI capabilities in AWS* LEX:
ASR [Automatic Speech Recognition] will translate Voice into Text.
NLU [Natural Language understanding] will understand the correct Intent from your Input text/phrase/language.
Designing a new AI Product using LEX. It can understand Voice/Speech/Text/language.
Your new LEX Bot can be easily integrated with all AWS services. This Bot can be deployed to any Bot platform — Slack, Facebook Messenger, etc.
We can use Voice and Text, communication channels to get work done. This is a Natural way of communication.
User and Bot — interaction workflow:
Voice commands->translate to Text->Understanding Language of Text
-> take Actions -> return Responses
NLU will try to understand your Input Text/Utterances and find the Correct Intent for your Utterance. If NLU does not find a Intent for your Input text, means your Bot is not working at optimum level.
**The Bot design concepts:
Intent — performs a specific Action
Utterances — Speech or Text phrases which specify your Intent
Slot — input fields required for the Intent
Fulfillment — take approval from User before taking Action
Action — API or DB call
Let’s understand in depth:
- Bot is your AI service, you will deploy to Facebook messenger and Slack
- Intent is what User wants to do ? commands for the Bot system
- Slot — input parameters for your Intent
- Attributes of an Slot = Slot name, Data Type[date, string, city, currency], Default values
- Bot is a collection of Intents
- Ur Intent will need collection of Slots for taking Action on behalf of User
- To take Action — the Bot will invoke the Lambda function and pass the validated Slot values taken from the User
- Finally, the response is returned to the User
- You can create Versions of your Bot, Intent and Slot
- We can Monitor our Bot