automate content moderation and

52

Upload: others

Post on 18-Apr-2022

3 views

Category:

Documents


0 download

TRANSCRIPT

Page 1: Automate content moderation and
Page 2: Automate content moderation and

© 2019, Amazon Web Services, Inc. or its affiliates. All rights reserved.

Automate content moderation and compliance with AI

A I M 3 0 5 - R

Kashif Imran

Sr. Solutions Architect, AWS AI

Amazon Web Services

Niranjan Hira

Product SA, AWS AI

Amazon Web Services

Gillian Armstrong

Solutions Engineer

Liberty Mutual

Page 3: Automate content moderation and

Agenda

Intros & expectations

Access your account

Images & video

Profanity detection with text

Page 4: Automate content moderation and

Related breakouts

AIM211 – AI document processing for business automation

AIM313 – Add computer vision to your applications

AIM206 – Finding a needle in a haystack: Use AI to transform content management

Page 5: Automate content moderation and

FRAMEWORKS INTERFACES INFRASTRUCTURE

Broadest and deepest set of capabilities

VISION SPEECH LANGUAGE CHATBOTS FORECASTING RECOMMENDATIONS

A M A Z O N

P O L L Y

A M A Z O N

T R A N S C R I B E

A M A Z O N

T R A N S L A T E

A M A Z O N

C O M P R E H E N D

& C O M P R E H E N D

M E D I C A L

A M A Z O N

L E X

A M A Z O N

F O R E C A S T

A M A Z O N

R E K O G N I T I O N

I M A G E

A M A Z O N

R E K O G N I T I O N

V I D E O

A M A Z O N

T E X T R A C TA M A Z O N

P E R S O N A L I Z E

Amazon SageMaker Ground Truth Notebooks Algorithms + Marketplace Reinforcement Learning Training Optimization Deployment Hosting

F P G A SE C 2 P 3

& P 3 D N

E C 2 G 4

E C 2 C 5

I N F E R E N T I AG R E E N G R A S S E L A S T I C

I N F E R E N C E

D L

C O N T A I N E R S

& A M I s

E L A S T I C

K U B E R N E T E S

S E R V I C E

E L A S T I C

C O N T A I N E R

S E R V I C E

The AWS ML stack

AI Services

ML Services

ML Frameworks + Infrastructure

Page 6: Automate content moderation and

Amazon Rekognition:Deep Learning-Based Image and Video Analysis

Page 7: Automate content moderation and

Amazon Rekognition Features

Faces Celebrities Labels

Moderation ScenesActivities Paths

Text

Page 8: Automate content moderation and

Amazon Rekognition Benefits

Low costYour data is your ownServerless

Rapid integration

State of the art capabilities

Continuous improvement

Page 9: Automate content moderation and

Amazon Textract

Text extraction Table extraction Form extraction

Page 10: Automate content moderation and

© 2019, Amazon Web Services, Inc. or its affiliates. All rights reserved.

Page 11: Automate content moderation and

Customer-facing instructions to access Event Engine

You can place these slides in your presentation.

You may need to copy/paste the contents of each slide, to keep your theme formatting and copyright footers intact.

Questions? Contact [email protected]

Page 12: Automate content moderation and

AWS Event Engine

https://dashboard.eventengine.run

The AWS Event Engine was

created to help AWS field teams

run Workshops, GameDays,

Bootcamps, Immersion Days,

and other events that require

hands-on access to AWS

accounts.

Page 13: Automate content moderation and

https://dashboard.eventengine.run

Page 14: Automate content moderation and

https://dashboard.eventengine.run

Page 15: Automate content moderation and

https://dashboard.eventengine.run/dashboard

Page 16: Automate content moderation and

https://dashboard.eventengine.run/dashboard

Page 17: Automate content moderation and

https://dashboard.eventengine.run/dashboard

Page 18: Automate content moderation and

https://dashboard.eventengine.run/dashboard

Page 19: Automate content moderation and

https://dashboard.eventengine.run/dashboard

Don’t forget to change the

permissions on the pem file.

For example:

chmod 400 ee-default-keypair.pem

Page 20: Automate content moderation and

© 2019, Amazon Web Services, Inc. or its affiliates. All rights reserved.

Page 21: Automate content moderation and

Amazon Comprehend

A fully managed and continuously trained service that discovers insights and relationships in text

Page 22: Automate content moderation and

Amazon Comprehend

Discover insights and relationships

in textEntities

Key Phrases

Language

Sentiment

Amazon

Comprehend

Topic Modeling

Page 23: Automate content moderation and

Bark & Co uses Amazon Comprehend to analyze

their customers’ comments and feedback,

allowing them to respond faster and improve the

customer experience.

Page 24: Automate content moderation and

Custom Amazon Comprehend

Page 25: Automate content moderation and

Custom Classification

• Triage support tickets

• Moderate forums

• Organize customer feedback

• Organize support calls

Classification:

“PRICING”

Classification:

“CANCEL_ACCOUNT”

Classification:

“LOYALTY_PROGRAM”

EXAMPLE

Page 26: Automate content moderation and

Automated Custom Classifier Training

Create a CSV file

with training data

Train the classifier Classify

TEXT LABEL

I am calling about my

credit card

LOYALTY_PROGRAM

I really need to shut the

service down

CANCEL_ACCOUNT

My points are not being

applied correctly

LOYALTY_PROGRAM

The service is very

expensive compared to

competition

PRICING

I’d subscribe if you had

discounts

PRICING

•Automated algorithm

selection

•Automated tuning and

testing

•SDK or code-free

Console UX

1 2 3

Page 27: Automate content moderation and

Custom Entities

• Analyze documents for your business and domain terms and

phrases

• Bring your own schema to unstructured text analytics

Hello, my name is John Doe, and thank you for calling

AnyCompany. I understand you are calling about part number

XT2457. We have that part on back order and we are expediting it

for you.

Thank you, but I was expediting it last week. At this point I think

we should go ahead and cancel the order entirely.

We are sorry to hear that sir, would you be willing to complete the

order if we offered a 10% discount?

Yes, thank you.

Person: John Doe

Organization: AnyCompany

Part: XT2457

Account_Action: cancel the order

Page 28: Automate content moderation and

Automated Custom Entity Recognizer Training

Prepare examples Train the recognizer1 2 3 Analyze

Terms

and

phrases

Documents

containing

terms and

phrases in

text

ENTITY VALUE ENTITY TYPE

XT2457 Part

AnyCompany Organization

John Doe Person

Cancel the order Account_Action •Automated annotation

•Automated algorithm

selection

•Automated tuning and

testing

•SDK or code-free

Console UX

Hello, my name is John Doe, and

thank you for calling

AnyCompany. I understand you

are calling about part number

XT2457. We have that part on

back order and we are expediting

it for you.

Thank you, but I was expediting it

last week. At this point I think we

should go ahead and cancel the

order entirely.

We are sorry to hear that sir,

would you be willing to complete

the order if we offered a 10%

discount?

Yes, thank you.

Page 29: Automate content moderation and

Amazon Comprehend Structured Output

Person Organization Part Account_

Action

John Doe AnyCompany XT2457 Cancel the

order

Richard Roe SomeCompany SD5578 Renew

subscription

Pre-trained entities Custom entities

Page 30: Automate content moderation and

Training data for Custom Entity Recognizer Training

Using entity list and

training docs1

Using annotations

and training docs2

• Simple and fast

• Specify entity values that are present in the

training documents

• May have poor accuracy if different entity

types have same values or values in wrong

context are selected for training

• Specify document offset for values

present in the training documents

• Higher annotation effort

• Higher accuracy as only specific values in

the desired context are selected for

training

OR

Page 31: Automate content moderation and

Let’s build

Offensive language

Hate speech

Profanity

Custom entity recognizer for profanity

Page 32: Automate content moderation and

© 2019, Amazon Web Services, Inc. or its affiliates. All rights reserved.

Page 33: Automate content moderation and

Using this out in the real world…

https://aws.amazon.com/solutions/media-analysis-solution/

Media Analysis Solution:

Page 34: Automate content moderation and

Web Front End

AI Analytics

Indexed for Searching

Amazon S3

Upload

Media

Search

Trigger

Metadata

Media Analysis Solution

Page 35: Automate content moderation and

Architecture Overview

AWS Elemental

MediaConvertAmazon

Transcribe

Amazon

Rekognition

Amazon

Comprehend

Amazon

Elasticsearch

Service

Amazon

CloudFront

Amazon

Cognito

Amazon API

GatewayAWS Lambda

AWS Step

FunctionsAmazon S3

(web interface)

Amazon S3

(Media Analysis

bucket)

AWS Lambda

Page 36: Automate content moderation and

We need to keep inappropriate content out of

our system!

Page 37: Automate content moderation and

We need to keep inappropriate content out of

our system!

No problem!

Page 38: Automate content moderation and

We need to keep inappropriate content out of

our system!

No problem!

Page 39: Automate content moderation and

Cost

Architecture

MonitoringCode

Testing

CI/CD

Security

Scalable

Page 40: Automate content moderation and

What’s the

Business Need

Page 41: Automate content moderation and

What’s the

Business Need

• What does inappropriate mean?

• What action will we take

if we find inappropriate

content?

• How accurate do we need it to be?

Page 42: Automate content moderation and

What’s the

Business Need

Page 43: Automate content moderation and

What’s the

Business Need

Action?? Monitoring

Human

in the loop

Data

Retention

Retraining

Identification

Page 44: Automate content moderation and

Amazon

Rekognition

Message shown to

customer advising

that the image

doesn’t meet our

guidelines.

Auto Claim Damage Photos Upload

Upload

Accept

Reject

Page 45: Automate content moderation and

Amazon

Rekognition

Auto Claim Damage Photos Upload

Upload

Accept

Reject

S3 Bucket

Page 46: Automate content moderation and

Customer Feedback Form

Amazon

Comprehend

Flag / UnflagAmazon Simple Queue

Service

AWS Lambda

CloudWatch

Event

AWS Step

Functions

Page 47: Automate content moderation and

Let’s Build…

Amazon

Comprehend

Flag / UnflagAmazon Simple Queue

Service

AWS Lambda

CloudWatch

Event

AWS Step

Functions

Page 48: Automate content moderation and

Yes

No

Trigger Step Function

Output

Page 49: Automate content moderation and

We need to keep inappropriate content out of

our system!

Sure! Let’s talk about what you

need.

Go Build

Something

Awesome

Page 50: Automate content moderation and

Yes

No

Trigger Step Function

Output

Page 51: Automate content moderation and

Thank you!

© 2019, Amazon Web Services, Inc. or its affiliates. All rights reserved.

Page 52: Automate content moderation and

© 2019, Amazon Web Services, Inc. or its affiliates. All rights reserved.