Text Analysis & Chatbots
Through Building

AI Trade School - Session 3B

How Machines Understand Language

🔄 Welcome Back

You've Built AI 6 Times Already

✍️

Week 1 Prompts

AI fundamentals mastered

📊

Week 2 Models

ML principles understood

👁️

Week 3A Vision

Computer Vision specialist

🗣️

Week 3B Today

NLP & Language AI

What is NLP?

Machines Finding Patterns
in Text

Just Like You Recognize Familiar Phrases Instantly

📊 Real Example: Sentiment Analysis

See What We're Building

INPUT: "This product is amazing! I love it!"

✓ POSITIVE (0.95 confidence)

INPUT: "Worst purchase ever. Don't buy this."

✗ NEGATIVE (0.98 confidence)

INPUT: "It's okay, nothing special."

○ NEUTRAL (0.60 confidence)

💬 Real Example: Chatbots

Q&A Pattern Recognition

You teach: "What are your hours?" → "9am-6pm, Mon-Fri"

User asks: "When are you open?" (slightly different wording)

Chatbot recognizes: It's the same INTENT as the first question

Chatbot responds: With the matching answer

That's intent recognition. That's NLP.

YOUR NLP JOURNEY

1

WEEK 1-2

Pattern Recognition

2

WEEK 3A

Visual Patterns

3

TODAY

Language Patterns

4

WEEK 4

Generate New Content

🚀 Career Story 1: Sarah

Customer Service Manager

Worked at call center, 8 hours/day reading complaints

Learned sentiment analysis to flag high-anger feedback

Built tool that identified customer emotions automatically

Now spends 2 hours on flagged complaints, 6 on strategy work

Promoted to Manager. Started with the skill you're learning TODAY.

🚀 Career Story 2: Marcus

Tech Support Engineer

Company received 500 support tickets per day

Built chatbot to automatically route tickets to right team

Used intent recognition to understand ticket category

Automated routing handled 80% of tickets without humans

Got raise + promotion to Senior Engineer. Built what you're building TODAY.

🚀 Career Story 3: Jamal

Startup Founder

Noticed companies paying expensive consultants for feedback analysis

Learned sentiment analysis and built customer feedback tool

Started selling it. Now processes 10M customer reviews/month

Multi-million dollar business from one skill

Started by building exactly what you're about to build TODAY.

🛤️ YOUR CHOICE: TWO PATHS

Path A: Sentiment Analyzer

Upload customer reviews or social media posts

Teach machine to understand: positive, negative, or neutral

Perfect if you love: Data, analytics, customer insights

Path B: Chatbot (FAQ Assistant)

Create Q&A pairs on topic you choose

Teach machine to find matching answers to new questions

Perfect if you love: Dialogue, creativity, conversation

Both equally valuable. Both will impress employers. Choose what excites YOU.

YOU CAN
DO THIS

Proof: You've Done It Before

Week 1 felt impossible. You figured it out.

Week 2 was too technical. You nailed it.

Week 3A (Computer Vision)? You mastered it.

Week 3B? Same pattern. Same outcome.

YOU'VE GOT THIS.

NOW WE BUILD

110 Minutes. Your NLP Application.

Setup

Choose path, access platform (10 min)

Practice

Guided demo with your platform (15 min)

Build

Create your artifact (70 min)

Polish

Finish & prepare to share (15 min)

🔑 Platform Access

Choose Your Tool

Path A (Sentiment): Azure Text Analytics OR AWS Comprehend

Path B (Chatbot): QnA Maker OR AWS Lex

All pre-configured. All ready to go. I'll walk you through each one.

📊 Guided Practice: Sentiment (Path A)

Follow Along on Your Laptop

Step 1: Log into your platform dashboard

Step 2: Paste a customer review into the text box

Step 3: Click "Analyze Sentiment"

Step 4: See the result (positive/negative/neutral + confidence score)

That's it. We'll do this together on screen. Everyone follow along.

💬 Guided Practice: Chatbot (Path B)

Follow Along on Your Laptop

Step 1: Log into your chatbot platform

Step 2: Create a new knowledge base (give it a name)

Step 3: Enter a Q&A pair (Question + Answer)

Step 4: Train the model (usually instant)

Step 5: Test it with a similar question

We'll go through this together. No rushing. Questions welcome.

✓ Readiness Check

Before We Build

Path A students: Can you paste text and see sentiment scores?

Path B students: Can you input Q&A pairs and train?

If your hand isn't up, flag me down now. We'll get you working before building starts.

🎯 Block 1 Goals (35 min)

Foundation Building

Path A

Analyze 15-20 customer reviews

Get sentiment score for each

Path B

Input 15-20 Q&A pairs

Train model + test

Then check in. We'll expand from there.

🎯 Block 2 Goals (35 min)

Adding Depth

Path A

Expand to 25-30 reviews

Find patterns: why positive? why negative?

Path B

Expand to 25-30 Q&A pairs

Test with 10+ diverse questions

Quality over quantity. Let's go deeper.

✓ Minimum Viable Artifact

What "Done" Looks Like

Path A

✓ 10+ reviews analyzed

✓ Can explain patterns

✓ Screenshot of results

Path B

✓ 10+ Q&A pairs trained

✓ 5+ test questions answered

✓ Screenshot of chatbot

⚠️ Common Issues & Fixes

We've Got You Covered

API not connecting: Check internet, use pre-configured account

Confused about sentiment scale: 0-0.4 negative, 0.4-0.6 neutral, 0.6-1.0 positive

Chatbot giving wrong answers: Your Q&A pairs need to be clearer

Laptop slow/freezing: Switch laptops, use backup device

Getting behind: Use simplified version, focus on quality over quantity

📊 Path A: Sentiment Deep Dive

Finding Patterns in Emotion

Look at high-scoring reviews: What words appear repeatedly?

Look at low-scoring reviews: What complaints show up?

What surprised you? Any unexpected scores?

If 10 reviews score positive and 2 score negative, why?

This analytical thinking is what makes this valuable to employers.

💬 Path B: Chatbot Deep Dive

Testing Intent Recognition

Test with questions SIMILAR to your Q&A pairs but worded differently

Does it find the right answer? That's intent recognition working.

If it fails, refine your Q&A pairs to be clearer

Accuracy improving = your training is working

This is literally how companies test chatbot quality in the real world.

SHOWCASE TIME

Celebrate Your Work

3-4 of you will share what you built

Mix of both paths (Sentiment & Chatbot)

Everyone listens and gives feedback

Peer validation is powerful. When your classmate says "that's cool," it means something.

🎯 What You Built Today

NLP Mastery

Taught a machine to understand language

Either: Analyzed sentiment in human text

Or: Built a conversational system

Tested it. Documented results. Understood why it works.

Built a real NLP application

THAT'S ADVANCED AI SKILL

You're Not Learning
ABOUT NLP

YOU ARE
NLP SPECIALISTS

7 artifacts built in 3 weeks

You understand WHEN to use each AI tool

You know HOW to build them

You can evaluate WHETHER it works

That's professional AI expertise right there.

Text Analysis & ChatbotsThrough Building

AI Trade School - Session 3B

🔄 Welcome Back

You've Built AI 6 Times Already

Week 1 Prompts

Week 2 Models

Week 3A Vision

Week 3B Today

What is NLP?

Machines Finding Patternsin Text

📊 Real Example: Sentiment Analysis

See What We're Building

💬 Real Example: Chatbots

Q&A Pattern Recognition

That's intent recognition. That's NLP.

YOUR NLP JOURNEY

🚀 Career Story 1: Sarah

Customer Service Manager

🚀 Career Story 2: Marcus

Tech Support Engineer

🚀 Career Story 3: Jamal

Startup Founder

🛤️ YOUR CHOICE: TWO PATHS

Path A: Sentiment Analyzer

Path B: Chatbot (FAQ Assistant)

Both equally valuable. Both will impress employers. Choose what excites YOU.

YOU CANDO THIS

Proof: You've Done It Before

YOU'VE GOT THIS.

NOW WE BUILD

110 Minutes. Your NLP Application.

Setup

Practice

Build

Polish

🔑 Platform Access

Choose Your Tool

📊 Guided Practice: Sentiment (Path A)

Follow Along on Your Laptop

💬 Guided Practice: Chatbot (Path B)

Follow Along on Your Laptop

✓ Readiness Check

Before We Build

🎯 Block 1 Goals (35 min)

Foundation Building

Path A

Path B

Then check in. We'll expand from there.

🎯 Block 2 Goals (35 min)

Adding Depth

Path A

Path B

Quality over quantity. Let's go deeper.

✓ Minimum Viable Artifact

What "Done" Looks Like

Path A

Path B

⚠️ Common Issues & Fixes

We've Got You Covered

📊 Path A: Sentiment Deep Dive

Finding Patterns in Emotion

💬 Path B: Chatbot Deep Dive

Testing Intent Recognition

SHOWCASE TIME

Celebrate Your Work

Peer validation is powerful. When your classmate says "that's cool," it means something.

🎯 What You Built Today

NLP Mastery

THAT'S ADVANCED AI SKILL

You're Not LearningABOUT NLP

YOU ARENLP SPECIALISTS

Text Analysis & Chatbots
Through Building

Machines Finding Patterns
in Text

YOU CAN
DO THIS

You're Not Learning
ABOUT NLP

YOU ARE
NLP SPECIALISTS