Intelligent Automation

101
 1 SESSION 1: RPA

TABLE
OF 2 SESSION 2: CHATBOT
CONTENTS
3 SESSION 3: COMPUTER VISION & NLP
Session 2
Artificial Intelligence

AI

Machine Learning/ Deep

Learning
Chatbot

Natural Language Processing

Virtual Assistant

Computer Vision
AI, Machine Learning & Deep Learning

Artificial Intelligence is the broader term which refers to development of theories and
AI algorithms which enables computer systems to perform tasks which require human
intelligence

AI

Machine learning is a branch of artificial intelligence which enables computer

Machine algorithms to learn from experience and data. Machine learning is primarily
Learning Machine categorized in three sub-sets:
Learning • Supervised learning
• Unsupervised learning
• Reinforcement learning
Deep
Learning
Deep learning is a subset of machine learning. Deep learning uses Artificial Neural
Network which is a series of connected nodes or layers to transmit and process data,
Deep
just like a connection of neurons in human brain. Deep learning uses massive amount
Learning
of data to be trained and generally requires more computational power to process data
than traditional machine learning algorithms
Machine Learning: Supervised Learning

Labeled Data

Prediction

Dog

Cat

Labels Model
training

Test Data
Cat

Dog
Gorilla
Machine Learning: Unsupervised Learning
Output

Input Raw Data

Unlabeled Data Dog

Cat

Model training and

processing

Bull
 Chatbot & Virtual Assistant

A chatbot is a software “Chatbots will fundamentally 70% white-collar $112B projected

application which revolutionize how computing is workers will interact with value of chatbot
simulates human-like experienced by everybody” – conversational platforms eCommerce transactions by
conversation with users Satya Nadella, Microsoft CEO daily by 2022 - Gartner 2023 – Juniper Research
via text message

Types of Chatbots & Assistants Examples of Assistants Major chatbot platforms

Rule Based – These chatbots use if/then Google Assistant

Google Dialog Flow
logic to create conversation flow

ML Based – More interactive, Siri

personalized chatbots/ assistants which Microsoft Bot
learn from data using machine learning/ Framework
deep learning algorithms Alexa

Hybrid – Combination of both rule Amazon Lex

Cortana
based and ML based approach
 The role and complexity of chatbots have evolved over the years

Types of Chatbots Why Chatbots?

Companion Counsels using emotion, humor and Millennials are more

empathy
comfortable in using
Level of Complexity

Understands user objectives and messaging app

Advisor provide recommendations

Performs tasks using API integration

Actor (approvals, process execution, update )
Reduced cost of
Identifies parts of speech and
Reader maintaining support staff
performs semantic parsing using NLP

Provides responses based on scripted

Assistant FAQs, keywords and / or a defined tree
structure
 Building blocks of chatbots

Q&A Intent and Entity Recognition

▪ Reply to user queries and requests through text, Understand ▪ Understand what the user is requesting and mapping
web links, audio and video it to the corresponding bot action
▪ Provide standard information, facts, solutions and ▪ Map natural language words to standard phrases or
details categories such as Christmas to 25 th December

Interact
Dialog Management Natural Language Processing
Analyze
▪ Track and remember user conversation history
▪ Parse natural human language text and voice in order to
over the same interaction and previous
understand unstructured inputs rather than just
interactions
depending upon FAQs or keywords
▪ Ask relevant follow up questions

Computation
Machine Learning
Learn ▪ Perform any calculations needed to fulfil the user
▪ Analyze human responses and learn how to requests or retrieve the required information
respond to the users ▪ Integrate with required applications, databases or
▪ Recognize qualitative intents channels through APIs

Reinforcement Learning Data Storage

▪ Learn from incorrect answers, accept
corrections and improve over a period of
time
▪ Store backend data needed to answer user requests
and queries
▪ Save user conversational history to use for future
conversations, learning from them and make needed
improvements over a period of time

CHATBOT FUNCTIONS
 Chatbot building using Google Dialogflow
Dialogflow is a natural language understanding platform used to design and integrate a conversational user interface into
mobile apps, web applications, devices, bots, interactive voice response systems and related uses. Dialogflow provides two
different virtual agent services, each of which has its own agent type, user interface, API, client

Dialogflow CX: Provides an advanced agent type suitable for large Dialogflow ES: Provides the standard agent type suitable for small
or very complex agents and simple agents

➢ 30+ languages and variants s ➢ 30+ languages and variants s

•Advanced performance dashboards •Performance dashboards
•Data export to custom dashboards •Data export to custom dashboards
•State-based visualizations •Integration into popular channels like Google Assistant, Slack,
•Integration across digital channels, including web, mobile, Twitter, etc.
messenger, etc. •One-click telephony integration
•Advanced one-click telephony integration •Form-based bot builder
•Reduce development time by 30% with an intuitive visual builder •Flat data model for simple use cases
for visual state machine •Supports up to 10 sub-agents with as many as 20,000 intents
•Reuse intents and intuitively define transitions and data •Intents and training phrases not shared across sub-agents
conditions
Models capable of easily defining and detecting minor
conversation detours
Supports settings like DTMF, live agent handoff, barge-in, speech
timeouts.
•Supports up to 20 independent conversation flows with 40,000
intents
•Shared intents and training phrases across flows
Google Dialogflow components
A Dialogflow agent is a virtual agent that handles concurrent conversations with your end-users. It is a natural language
understanding module that understands the nuances of human language. Dialogflow translates end-user text or audio
1 Agent during a conversation to structured data that your apps and services can understand. You design and build a Dialogflow
agent to handle the types of conversations required for your system.

An intent categorizes an end-user's intention for one conversation turn. For each agent, you define many intents, where
your combined intents can handle a complete conversation. When an end-user writes or says something, referred to as
2 Intent an end-user expression, Dialogflow matches the end-user expression to the best intent in your agent. Matching an intent
is also known as intent classification

Each intent parameter has a type, called the entity type, which dictates exactly how data from an end-user expression is
extracted. Dialogflow provides predefined system entities that can match many common types of data. For example,
3 Entity there are system entities for matching dates, times, colors, email addresses, and so on. You can also create your
own custom entities for matching custom data. For example, you could define a vegetable entity that can match the
types of vegetables available for purchase with a grocery store agent.

Dialogflow contexts are similar to natural language context. If a person says to you "they are orange", you need context in
order to understand what "they" is referring to. Similarly, for Dialogflow to handle an end-user expression like that, it
needs to be provided with context in order to correctly match an intent.
4 Context Using contexts, you can control the flow of a conversation. You can configure contexts for an intent by setting input and
output contexts, which are identified by string names. When an intent is matched, any configuredoutput contexts for
that intent become active. While any contexts are active, Dialogflow is more likely to match intents that are configured
with input contexts that correspond to the currently active contexts.
 Google Dialogflow architecture
The main platforms that are compatible with Dialogflow are listed below. It can be integrated to all of the basic conversation
platforms, except WhatsApp. Using other platforms such as Twilio, Dialogflow can be integrated into WhatsApp, too.

• Google Assistant
• Slack
• Twitter
• Twilio
• Skype
• Telegram
• Facebook Messenger
• Amazon Alexa
 Algorithm used in Dialogflow
Intent matching
When an end-user writes or says something, referred to as an end-user expression, Dialogflow compares
the expression to the training phrases for each intent to find the best match. Matching an intent is also
known as intent classification. This document describes the factors used to match an intent.

Matching algorithms
Dialogflow uses two algorithms to match intents: rule-based grammar matching and ML matching.
Dialogflow simultaneously attempts both algorithms and chooses the best result.
 Chatbot examples

Examples of chatbot applications

1 A chatbot for ordering food that allows customers to order from their office or home

2 A chatbot that answers customer service questions

3 A chatbot that allows the customer to book a flight and receive relevant information

4 A chatbot that helps a customer to purchase in eCommerce platforms

5 A health bot that provides services for booking a Doctor consultation and getting remedies for different symptoms

6 A chatbot that books movie tickets and also provided the reviews
 Chatbot use case

Use case 1: Voicebots for customer service

Use case 2: Chatbots for B2C conversations

Thank You