SlideShare a Scribd company logo
Introducing Amazon
Mechanical Turk
A Brief Overview of What, Why, and How
John Hoskins
Agenda
What is Mechanical Turk?
When would you use it?
How are others using it - Case Studies
How – Implementation options
What is
Mechanical Turk?
Mechanical Turk is a
   Marketplace.
Mechanical Turk is a
         marketplace for work.
Mechanical Turk gives businesses and developers access
to an on-demand, scalable workforce.
  •   Flexibility: Scale your workforce up and down quickly
  •   Accuracy: Get high quality, efficient and cost effective results.
  •   Price: Pay only when you are satisfied with the results.
  •   Speed: Start receiving results in minutes
How it works:.
“Design & Publish”      “Validate,
                        Pay & Go”
Workforce
500,000 Workers
190+ Countries
Who are the Workers?
•   Workers are global
•   Workers represent numerous skill sets
•   24X7 Follow the sun availability

Managing Your Workforce
•   You can choose who is allowed to work on your assignments
•   Can narrow to US based (i.e. when you need western culture competence)
•   Leverage qualifications to find your best Workers
When would you use
 Mechanical Turk?
Business Challenges
Our Community is generating 10,000 comments per day
– how can we eliminate SPAM?
We pull information from hundreds of sources – how do
we normalize and cleanse?
We have thousands of legacy assets – how do we make
them searchable?
Our heuristics need jump started – how can I generate
thousands of reference data points fast?
We need to add a new attribute to our data?
Business Challenges
Requires Human Judgment: Recognition, Reasoning,
Context, Prior Knowledge

Cannot be completely solved algorithmically

Has unpredictable fluctuations in volume
How are businesses using it?
   Data Management                     Categorization
       Data Verification                   Classification
       Data Entry & Collection             Tagging
       Data De-duplication                 Keywords
       Algorithm Training                  Sentiment Analysis


   Content & Media                     Business Services
       Moderate Photos & Content           Search Relevancy
       Content Creation & Editing          Data Gathering
       Transcription
                                            Enhancement
How are others using
 Mechanical Turk?
Search Enhancement
Make your assets more discoverable
• Categorize them into your search tree
• Add descriptions, attributes and keywords
• Enhance listing data


How it works:
Item description
                     Workers select
    is sent to                          Items are placed     Content is more
                    the best category
Mechanical Turk                         in that branch of   easily discovered
                    and sub-category
  for review by                          the search tree     through search
                       for the item
     Workers
Data Handling
Resolve Data conflicts directly in your business process
•   Remove duplicates
•   Normalize data entries
•   Resolve conflicts and collisions
•   Enhance entries

How it works:
   Business
                    Items flagged for
Process Ingests                           Workers review    Business
                     conflict are sent
Data Elements                            item and resolve    Process
                      to Mechanical
 from multiple                                conflict      continues
                           Turk
    sources
How to apply your task to
   Mechanical Turk?
Your Project  Tasks
“Working Backwards”
•   What are the business results that you want?
•   How will you use those results?
•   What format do the results need to be in?
•   Which tasks benefits from human intelligence?

Break your overall project into smaller “tasks”
• Consolidate like items and processes

Define instructions for each task
Example Workflow
Content Moderation with API integration
The Key components of your HIT
Instruction
• Guidelines for how to do
  a HIT


Data                         What kind of dress is this?
• “batch” concept                 Cocktail

                                  Sun dress

Question(s)                       Bridal dress
Work Catalog
               Workers
               discover
               available HITs
               through the
               listings catalog
               Name of
               Requester, Na
               me of
               HIT, Price and
               Qualifications
               are displayed
HIT Example
              Specific
              instructions on
              what to do –
              and how to do it
              Providing as
              much
              information as
              available
Adjudication Strategies
Leverage different methods of determining the correct answer
by designing for accuracy in your HIT:
• Multiple Choice
    • Plurality
    • Known Answers (“Gold Standards”)
• Free-form
    • Multiple HIT Workflow’s where a new Worker “edits” or “grades”
      previous Worker submissions

• Creative Tasks
    • Bonus Payments
Your Workers
Identify and manage your best Workers
                           • All available Workers
             All Workers   • 500,000 +
                           • 24x7, 365

                           • Any Worker who
                             completes a HIT
            Your Workers   • Evaluate performance on
                             your HITS

            Your Trusted   • Your best Workers with
                             proven quality and
              Workers        availability
Balances & Levers
Leverage market dynamics to optimize speed, price and accuracy to
meet your business needs.


                             Speed



                     Price

                                Accuracy
Requester access:
                                        Apps & Solution
     Web UI              API & CLT
                                          Providers
• Easy Access        • Programmatic    • Experts at
• Create HITS in       Access            Workflow design
  HTML               • Multiple SDKS   • They do the
• Upload and         • Massive scale     training and
  manage data via                        management of
  CSV                                    Workers for
                                         specific kinds of
                                         tasks
Self Service Applications
Mechanical Turk supports a robust ecosystem of providers
that have built or can build solutions to meet your needs.
Custom Solution Providers
Advantages of Mechanical Turk
                  Gives you access to an on-demand workforce
  Eliminates      100% Efficiency
Staffing Issues   Maintain the flexibility your business demands
                  Use only the capacity you need, when you need it
    Lowers        Converts a fixed cost into a variable cost
     Costs        Significantly reduces ongoing headcount expenses
                  Offers a pay-as-you-go staffing model
                  Eliminates lead time of interviewing and staffing
 Reduces work
                  Work gets done faster, by working in parallel
turnaround time
                  Work continues around the globe, around the clock
Thank you.
John Hoskins
hoskins@amazon.com
206.266.8105

     @amazonmturk

     Facebook.com/amazonmturk
     mechanicalturk.typepad.com

More Related Content

PPTX
Introduction to Amazon Mechanical Turk
PPTX
Sucessful implementation of JIRA and Confluence - tips and best practice
PPTX
Jira software 8.0 8.5 community presentation
PPTX
Introduction to JIRA
PPTX
An Overview of Robotic Process Automation (RPA)
PDF
Introduction to RPA and Document Understanding
PPTX
RPA Uipath Presentation.pptx
PPSX
JIRA
Introduction to Amazon Mechanical Turk
Sucessful implementation of JIRA and Confluence - tips and best practice
Jira software 8.0 8.5 community presentation
Introduction to JIRA
An Overview of Robotic Process Automation (RPA)
Introduction to RPA and Document Understanding
RPA Uipath Presentation.pptx
JIRA

What's hot (12)

PPS
Agile Project Management with Scrum
PDF
Robotic Process Automation-RPA
PPT
Executive Presentation on Agile Project Management by Boardroom Metrics Inc.
PDF
3 Amazing Examples of Real-life RPA Use Cases - Signity
PPT
Test automation workshop slideset
PDF
On Call Process (PDF)
PPTX
Software Agility.pptx
PDF
PPTX
PPT
Automated Testing vs Manual Testing
PPT
Introduction To Jira Slide Share
PPTX
UiPath Test Suite
Agile Project Management with Scrum
Robotic Process Automation-RPA
Executive Presentation on Agile Project Management by Boardroom Metrics Inc.
3 Amazing Examples of Real-life RPA Use Cases - Signity
Test automation workshop slideset
On Call Process (PDF)
Software Agility.pptx
Automated Testing vs Manual Testing
Introduction To Jira Slide Share
UiPath Test Suite
Ad

Viewers also liked (20)

PPTX
Amazon Mechanical Turk: Intro & Overview
PPTX
HRM in Coca Cola
PPTX
HRM process of coca cola beverages of pakistan ltd.
PDF
Amazon.com - Company Analysis (OD & HRM)
PDF
Amazon.com Business Model
DOCX
Amazon - Brand Management
PPTX
Eicosanoids - power point presentaion
DOCX
Strategic human resource management at COCA COLA BEVERAGES PAKISTAN LIMITED
PPT
Human resource management within coca cola
PPTX
Presentation on Amazon
PPTX
HR presentation oc COCA COLA
PPTX
Amazon: Strategic Management
PPTX
Hr policies at coca cola
PPTX
Company presentation - Amazon
PPTX
Training and development slides (2)
PDF
Training and development ppt
PPTX
Amazon ppt
PDF
Amazon.com Strategic Analysis
PPTX
training & development ppt
PPTX
Amazon Business Model
Amazon Mechanical Turk: Intro & Overview
HRM in Coca Cola
HRM process of coca cola beverages of pakistan ltd.
Amazon.com - Company Analysis (OD & HRM)
Amazon.com Business Model
Amazon - Brand Management
Eicosanoids - power point presentaion
Strategic human resource management at COCA COLA BEVERAGES PAKISTAN LIMITED
Human resource management within coca cola
Presentation on Amazon
HR presentation oc COCA COLA
Amazon: Strategic Management
Hr policies at coca cola
Company presentation - Amazon
Training and development slides (2)
Training and development ppt
Amazon ppt
Amazon.com Strategic Analysis
training & development ppt
Amazon Business Model
Ad

Similar to Amazon Mechanical Turk Overview (20)

PPTX
The ProPricer Proven Performance
PPTX
General Presentation
PPTX
Dashlane Mission Teams
PPTX
Social Media Use Cases - Webinar
PDF
xRM4Legal 2011 in the Cloud
PPTX
Behind the Curtain: Real-world HR Tech Implementations and What You Need to ...
PPTX
How to plan your Modern Workplace Project - SPS Denver October 2018
PPTX
Re-engineered Job Postings
PDF
Now that you've sold it how do you build it - XMPie Users Conference XUG 202...
PPSX
Talent Centric Presentation
PPTX
HR Cloud Cleanup: How To Effectively Connect HR Cloud Applications To Save Ti...
PPTX
De-Mystifying ROI Calculation for SharePoint
PPTX
Field report: Rapid application development
PDF
About Us | Taction Software
PDF
Talent Alpha: Unleashing talent of the global tech workforce
PDF
Optimizing Innovation: Modular Toolchains that Enable Digital Transformations
PDF
Optimizing Innovation- Modular Toolchains that Enable Digital Transformations
PPTX
Scaling Training Data for AI Applications
PDF
Service Proposition
PPTX
Amazon Web Services - Value Chain Analysis
The ProPricer Proven Performance
General Presentation
Dashlane Mission Teams
Social Media Use Cases - Webinar
xRM4Legal 2011 in the Cloud
Behind the Curtain: Real-world HR Tech Implementations and What You Need to ...
How to plan your Modern Workplace Project - SPS Denver October 2018
Re-engineered Job Postings
Now that you've sold it how do you build it - XMPie Users Conference XUG 202...
Talent Centric Presentation
HR Cloud Cleanup: How To Effectively Connect HR Cloud Applications To Save Ti...
De-Mystifying ROI Calculation for SharePoint
Field report: Rapid application development
About Us | Taction Software
Talent Alpha: Unleashing talent of the global tech workforce
Optimizing Innovation: Modular Toolchains that Enable Digital Transformations
Optimizing Innovation- Modular Toolchains that Enable Digital Transformations
Scaling Training Data for AI Applications
Service Proposition
Amazon Web Services - Value Chain Analysis

Recently uploaded (20)

PDF
Per capita expenditure prediction using model stacking based on satellite ima...
PDF
Building Integrated photovoltaic BIPV_UPV.pdf
PDF
Build a system with the filesystem maintained by OSTree @ COSCUP 2025
PDF
NewMind AI Weekly Chronicles - August'25 Week I
PPTX
A Presentation on Artificial Intelligence
PDF
Review of recent advances in non-invasive hemoglobin estimation
PDF
The Rise and Fall of 3GPP – Time for a Sabbatical?
PPTX
VMware vSphere Foundation How to Sell Presentation-Ver1.4-2-14-2024.pptx
PDF
Unlocking AI with Model Context Protocol (MCP)
PDF
Machine learning based COVID-19 study performance prediction
PDF
TokAI - TikTok AI Agent : The First AI Application That Analyzes 10,000+ Vira...
PDF
Peak of Data & AI Encore- AI for Metadata and Smarter Workflows
PPTX
Cloud computing and distributed systems.
PDF
Dropbox Q2 2025 Financial Results & Investor Presentation
PDF
NewMind AI Monthly Chronicles - July 2025
PDF
Chapter 3 Spatial Domain Image Processing.pdf
DOCX
The AUB Centre for AI in Media Proposal.docx
PDF
Electronic commerce courselecture one. Pdf
PDF
Shreyas Phanse Resume: Experienced Backend Engineer | Java • Spring Boot • Ka...
PDF
Reach Out and Touch Someone: Haptics and Empathic Computing
Per capita expenditure prediction using model stacking based on satellite ima...
Building Integrated photovoltaic BIPV_UPV.pdf
Build a system with the filesystem maintained by OSTree @ COSCUP 2025
NewMind AI Weekly Chronicles - August'25 Week I
A Presentation on Artificial Intelligence
Review of recent advances in non-invasive hemoglobin estimation
The Rise and Fall of 3GPP – Time for a Sabbatical?
VMware vSphere Foundation How to Sell Presentation-Ver1.4-2-14-2024.pptx
Unlocking AI with Model Context Protocol (MCP)
Machine learning based COVID-19 study performance prediction
TokAI - TikTok AI Agent : The First AI Application That Analyzes 10,000+ Vira...
Peak of Data & AI Encore- AI for Metadata and Smarter Workflows
Cloud computing and distributed systems.
Dropbox Q2 2025 Financial Results & Investor Presentation
NewMind AI Monthly Chronicles - July 2025
Chapter 3 Spatial Domain Image Processing.pdf
The AUB Centre for AI in Media Proposal.docx
Electronic commerce courselecture one. Pdf
Shreyas Phanse Resume: Experienced Backend Engineer | Java • Spring Boot • Ka...
Reach Out and Touch Someone: Haptics and Empathic Computing

Amazon Mechanical Turk Overview

  • 1. Introducing Amazon Mechanical Turk A Brief Overview of What, Why, and How John Hoskins
  • 2. Agenda What is Mechanical Turk? When would you use it? How are others using it - Case Studies How – Implementation options
  • 4. Mechanical Turk is a Marketplace.
  • 5. Mechanical Turk is a marketplace for work. Mechanical Turk gives businesses and developers access to an on-demand, scalable workforce. • Flexibility: Scale your workforce up and down quickly • Accuracy: Get high quality, efficient and cost effective results. • Price: Pay only when you are satisfied with the results. • Speed: Start receiving results in minutes
  • 6. How it works:. “Design & Publish” “Validate, Pay & Go”
  • 7. Workforce 500,000 Workers 190+ Countries Who are the Workers? • Workers are global • Workers represent numerous skill sets • 24X7 Follow the sun availability Managing Your Workforce • You can choose who is allowed to work on your assignments • Can narrow to US based (i.e. when you need western culture competence) • Leverage qualifications to find your best Workers
  • 8. When would you use Mechanical Turk?
  • 9. Business Challenges Our Community is generating 10,000 comments per day – how can we eliminate SPAM? We pull information from hundreds of sources – how do we normalize and cleanse? We have thousands of legacy assets – how do we make them searchable? Our heuristics need jump started – how can I generate thousands of reference data points fast? We need to add a new attribute to our data?
  • 10. Business Challenges Requires Human Judgment: Recognition, Reasoning, Context, Prior Knowledge Cannot be completely solved algorithmically Has unpredictable fluctuations in volume
  • 11. How are businesses using it?  Data Management  Categorization  Data Verification  Classification  Data Entry & Collection  Tagging  Data De-duplication  Keywords  Algorithm Training  Sentiment Analysis  Content & Media  Business Services  Moderate Photos & Content  Search Relevancy  Content Creation & Editing  Data Gathering  Transcription  Enhancement
  • 12. How are others using Mechanical Turk?
  • 13. Search Enhancement Make your assets more discoverable • Categorize them into your search tree • Add descriptions, attributes and keywords • Enhance listing data How it works: Item description Workers select is sent to Items are placed Content is more the best category Mechanical Turk in that branch of easily discovered and sub-category for review by the search tree through search for the item Workers
  • 14. Data Handling Resolve Data conflicts directly in your business process • Remove duplicates • Normalize data entries • Resolve conflicts and collisions • Enhance entries How it works: Business Items flagged for Process Ingests Workers review Business conflict are sent Data Elements item and resolve Process to Mechanical from multiple conflict continues Turk sources
  • 15. How to apply your task to Mechanical Turk?
  • 16. Your Project  Tasks “Working Backwards” • What are the business results that you want? • How will you use those results? • What format do the results need to be in? • Which tasks benefits from human intelligence? Break your overall project into smaller “tasks” • Consolidate like items and processes Define instructions for each task
  • 17. Example Workflow Content Moderation with API integration
  • 18. The Key components of your HIT Instruction • Guidelines for how to do a HIT Data What kind of dress is this? • “batch” concept Cocktail Sun dress Question(s) Bridal dress
  • 19. Work Catalog Workers discover available HITs through the listings catalog Name of Requester, Na me of HIT, Price and Qualifications are displayed
  • 20. HIT Example Specific instructions on what to do – and how to do it Providing as much information as available
  • 21. Adjudication Strategies Leverage different methods of determining the correct answer by designing for accuracy in your HIT: • Multiple Choice • Plurality • Known Answers (“Gold Standards”) • Free-form • Multiple HIT Workflow’s where a new Worker “edits” or “grades” previous Worker submissions • Creative Tasks • Bonus Payments
  • 22. Your Workers Identify and manage your best Workers • All available Workers All Workers • 500,000 + • 24x7, 365 • Any Worker who completes a HIT Your Workers • Evaluate performance on your HITS Your Trusted • Your best Workers with proven quality and Workers availability
  • 23. Balances & Levers Leverage market dynamics to optimize speed, price and accuracy to meet your business needs. Speed Price Accuracy
  • 24. Requester access: Apps & Solution Web UI API & CLT Providers • Easy Access • Programmatic • Experts at • Create HITS in Access Workflow design HTML • Multiple SDKS • They do the • Upload and • Massive scale training and manage data via management of CSV Workers for specific kinds of tasks
  • 25. Self Service Applications Mechanical Turk supports a robust ecosystem of providers that have built or can build solutions to meet your needs.
  • 27. Advantages of Mechanical Turk Gives you access to an on-demand workforce Eliminates 100% Efficiency Staffing Issues Maintain the flexibility your business demands Use only the capacity you need, when you need it Lowers Converts a fixed cost into a variable cost Costs Significantly reduces ongoing headcount expenses Offers a pay-as-you-go staffing model Eliminates lead time of interviewing and staffing Reduces work Work gets done faster, by working in parallel turnaround time Work continues around the globe, around the clock
  • 28. Thank you. John Hoskins hoskins@amazon.com 206.266.8105 @amazonmturk Facebook.com/amazonmturk mechanicalturk.typepad.com

Editor's Notes

  • #6: Amazon Mechanical TurkIt is a marketplace for work. In a traditional marketplace, like the Amazon.com retail site that you’re probably already familiar with, there are buyers who spend money to purchase products and there are sellers who make money by selling their products. This is true in the Amazon Mechanical Turk Marketplace as well. Businesses, or Requesters as they are called on Mechanical Turk are creating an offer for work. Mechanical Turk Workers accept and execute the offer, sell their work, and earn money when the transaction is complete. And just like in the Amazon.com retail experience, Amazon Mechanical Turk takes care of all of the transaction details including getting the work “listed” on the site so Workers see it when they browse for Work as well as completing the payment transaction when work is submitted and approved by you, the Requester. Let’s talk in a bit more detail about how it works.
  • #7: Requesters design HITs, Human intelligence tasks. These HITs are composed of Assignments which are published to the Mechanical Turk Marketplace catalog. Workers search for and select tasks to complete and submit their assignments to Requesters.Requesters approve or reject the results and pay Workers.
  • #8: Think of the Workers like customers. They are a large group of individuals, conditioned to come to the marketplace and search for work on a regular basis. The scale and scope of this group is what makes it powerful.
  • #11: Mechanical Turk is ideal for tasks that require a human assessment – things that require recognition or reasoning. Context and prior knowledge.Many of our customers use Mechanical Turk in conjunction with an algorithmic process – in areas where the process can’t make a judgment with high confidence – the 80/20 rule where the 20% that doesn’t have a clear answer is sent to a human for resolution.Mechanical Turk, like all of Amazon’s web service is scalable – in both directions, scale up to meet peaks, and back off when resources aren’t needed. Avoid paying for idle work capacity to meet peak demand.
  • #14: A large media company has an archive of photos with very cryptic descriptions. They need to categorize them into a taxonomy that helps customers find what they are looking for easily. This requires that a human look at every photo and place it into the proper category. The company sends the photo to Mechanical Turk where a worker views it and places it into the proper category and sub-category. Is it a landscape, person, object, etc.
  • #15: A large business listing service receives daily data feeds from many sources. In order to apply that data to the proper business, they have to resolve which business the information belongs and in what sequence. Their internal process is able to resolve approximately 80% of the incoming data – what do they do for the other 20%? They create a task on Mechanical Turk to have a human look at the information and make a judgment on its resolution. Given the answer – the ingestion process continues.
  • #17: Begin with a projectDefine the goals and key components of your project. For example, your goal might be to clean your business listing database so that you have accurate information for consumers. The sub-components of your project might be to categorize the businesses by listing type (i.e., restaurant or service) and verify that the related address and phone number are current. Break it into tasks and design your HITBreak the project into individual tasks so that many Workers can work in parallel to get your project done faster. For example, if you have 1,000 listings to verify, each listing would be an individual task.Next, design your Human Intelligence Tasks (HITs) by writing crisp and clear instructions, identifying the specific outputs and inputs desired and how much you will pay to have work completed.
  • #28: Mechanical Turk offers a business process that ensures 100% worker productivity and keeping the cost per submitted work order, (HIT) to the lowest level. Mturk relieves businesses from having to staff to peaks and realizing employee downtime as demand lessens. Businesses avoid the fully burdened costs of a permanent workforce. Also, a workforce that performs at a 70-75% productivity range.