ChatBot is a computer program that simulate human conversation, through artificial intelligence.
Some Famous Chatbots
- 1966 ELIZA
- 2015 ALEXA
- 2016 TAY
Why?
- More Secure for privacy
- Less Latency for requests
- Faster than Human Resposne
- Flexibilty
- Personalized Experience
- Easy Customer Suport and Service
Archirecture-
| Sensory Layer | <--> | ChatBotPlatforms | <--> | API Brokerage Layer |
|
|
(NLP and ML modules)
- Each chatbots has a set of intent which shows intentions of user.
- Suppose a pizza delivery chatbot has 2 intent i.e. order and cancel
- Suppose a banking chatbot has following intents i.e. check balance, change pin, transfer money
- Based on input chatbot select a particular intent
- Additional information user has to provide. Can be called as constraints.
- Suppose order a pizza has 2 slots i.e. what kind of pizza and size of pizza
- Change my pin requires current pin.
- Questions to be asked to get required slots.
- Makes Chatbots intelligent,
- identify context of chat.
- Like in change pin after entering current pin it should change pin.
- Actual work we want chatbot to do.
- In case of pizza delivery, request must be sent to restaurant by chatbot.
- request is sent to API brokerage layer.
- AWS Lex as a Chatbot platform
- AWS Lambda as a compute service.
- AWS DynamoDB as a NoSQL database.
https://exadatum-mit.signin.aws.amazon.com/console
Username :- registered email
Password :- exadatum-mit
-
create a custom bot
-
Session timeout - how much time it will wait for user response from prev context
-
IAM role - set of policies ( for security, permissions )
-
Create Intent
-
give some utterances.
-
give fulfillments
-
In lambda initialization we can assign some validation checks, database connection etc.
- Create a lambda function and use Oxford dictionary to get meaning.
- integrate a bot with messenger