SlideShare uma empresa Scribd logo
1 de 71
Jeff Rzeszotarski
[rez-oh-tar’-ski]
4th Year Ph.D. Student
HCII, CMU
www.jeffrz.com
jeffrz@cs.cmu.edu
Instrumenting the Crowd
James Cridland @ Flickr
James Cridland @ Flickr
James Cridland @ Flickr
Amazon Mechanical Turk
oDesk
CrowdFlower
Bernstein et al. 2010
Task Design
Gold Standard
Worker Agreement
Task Design
Cheating Resistant Design
Kittur, Chi, Suh, 2008
Von Ahn & Dabbish, 2004
Gold Standards
Gold Standard Worker 1 Worker 2
A A A
D B D
B C
A A
2 4 2
2 8
Callison-Burch, 2009
Downs et al. 2010
Worker Agreement
Callison-Burch, 2009 Ipeirotis et al. 2010
Dekel & Shamir, 2009 Snow et al. 2008
WorkerID Video Transcript
AF34DB9 Coloring comics in Photoshop is easy. First you…
QR40Q2 <blank>
TK421S9S Coloring comcs in photoshop is easy first u…
WOPR21 QOJEORJOFJDSKGNDKSNGSDSJGKDSJGLKSDJGL
APC3C55 Coloring comics in Photoshop is easy. First you…
IDKFAD1 Coloring comics in photoshop is easy. First you…
Disadvantages
Task Design
- Task-dependent
- Time consuming iteration
Gold Standard
- Not always available
- Not always applicable
Worker Agreement
- Limits response range
- Subject to majority effects
- Not always applicable
Worker A Worker B
Worker A
Worker A
Worker A
Worker A
Worker A
Worker B
Worker B
Worker B
Worker B
Worker B
Worker B
Worker B
Logging
ABC
SCROLL
MOUSE
MOVE
MOUSE
MOVE CLICK CLICKDELAY DELAYTYPING
120px 300px,
20px
Field1 1.1s ‘Hello’ 0.9s 100px,
220px
Submit
Example Log: Typing and submitting “Hello”
Features
Experimental MTurk Tasks
Noun Classification
“Identify which words are nouns.”
Image Tagging
“Generate 3-5 tags for each of 4 images.”
Reading Comprehension
“Read and then answer the following questions.”
Image Tagging
Provide at least 3 tags for
the following image…
ocean mountains
pretty
x4
Image Tagging
Image Tagging
52 participants, 114 submissions
Human Labeled Cheating
17 cheats, 97 regular
Human Labeled Quality
AVG 3.5 / 5, SD 1.13
Model Features
# Unique Letters
Total Time Elap.
Image Tagging
17 cheats, 97 regular
decision tree, 10-fold CV
93.0% accuracy
Image Tagging
Regressing on ‘quality’
SMOreg, 10-fold CV
r=0.5874 (generally ±0.5 on Likert)
Model Features
# Unique Letters
Total Time Elapsed
# of Clicks
Fields Accessed
Generalizing Models
Training
Reading
Comprehension
63 points
Testing
Noun
Identification
185 Points
Model predicts for a similar task
r=0.4948
Other Findings
Clustering workers by
behavioral patterns
De-anonymizing
worker submissions
Hanging Questions
Complex & creative work?
Human interpretability
Organizer cognitive models
CrowdScape
ABC
SCROLL
MOUSE
MOVE
MOUSE
MOVE CLICK CLICKDELAY DELAYTYPING
120px 300px,
20px
Field1 1.1s ‘Hello’ 0.9s 100px,
220px
Submit
Interac(vely	
  visualizing	
  user	
  
behavior	
  and	
  output
Thanks, d3 toolkit
Series and series of works starting with “Godzilla"
monster movie Godzilla was released in special
effects is the name of the fictional monster of
their work.
The film named “Godzilla" is a special effects
monster movie, and Godzilla is the monster.
ゴジラはに公開した特撮怪獣映画『ゴジラ』に始まる一連のシリーズ
作品及び、それらの作品に登場する架空の怪獣の名称である。	
  
I like to visit Alappuzha in Kerala, the God's own country.
This place is blessed with heavenly lakes. The view of lake
shores from the boats is very beautiful. We can enjoy cool
wind standing in the boats. We can see a complete
picture of a beautiful village by a trip through the lakes of
Alappuzha.
My favorite place in the world to be is right at home!
Ideally my daughter and husband and I just spending
time together watching television, sharing a meal, or
discussing our day. I know it might sound boring to some
but I find joy and security in the familiar surroundings. Sure,
there are other places I enjoy but given the choice my
home will always be the place I am the happiest.
Have you ever wondered what causes swing sets to move
with no one on them? Or why bridges twist and tear
themselves apart? The answer is aeroelastic tension. But the
science behind it is the most interesting part. Watch this and
you'll find out how it works in less than two minutes.
collapse bridge physics tension aeroelastic
Tacoma Narrows Bridge Collapse
Limitations
Logging is not perfect
Information overload
Worker traces
Parallel coordinates
Interpreting behavior
Feature set, ML algorithms

Mais conteúdo relacionado

Destaque

12 месяцев года ОБЩИЕ ТЕНДЕНЦИИ 2010 года
12 месяцев года ОБЩИЕ  ТЕНДЕНЦИИ 2010 года12 месяцев года ОБЩИЕ  ТЕНДЕНЦИИ 2010 года
12 месяцев года ОБЩИЕ ТЕНДЕНЦИИ 2010 года
ATOR
 

Destaque (12)

Презентация Василия Кузнецова, SweetCard: «Таргетированные предложения скидок...
Презентация Василия Кузнецова, SweetCard: «Таргетированные предложения скидок...Презентация Василия Кузнецова, SweetCard: «Таргетированные предложения скидок...
Презентация Василия Кузнецова, SweetCard: «Таргетированные предложения скидок...
 
Ux & Marketing - Meetup Flupa Toulouse
Ux & Marketing - Meetup Flupa Toulouse Ux & Marketing - Meetup Flupa Toulouse
Ux & Marketing - Meetup Flupa Toulouse
 
Анализ поисковых запросов в системе сравнения цен на туры Слетать.ру за июль ...
Анализ поисковых запросов в системе сравнения цен на туры Слетать.ру за июль ...Анализ поисковых запросов в системе сравнения цен на туры Слетать.ру за июль ...
Анализ поисковых запросов в системе сравнения цен на туры Слетать.ру за июль ...
 
Curiculum Vitae (CV)
Curiculum Vitae (CV)Curiculum Vitae (CV)
Curiculum Vitae (CV)
 
12 месяцев года ОБЩИЕ ТЕНДЕНЦИИ 2010 года
12 месяцев года ОБЩИЕ  ТЕНДЕНЦИИ 2010 года12 месяцев года ОБЩИЕ  ТЕНДЕНЦИИ 2010 года
12 месяцев года ОБЩИЕ ТЕНДЕНЦИИ 2010 года
 
19. Lead Management: far maturare i lead
19. Lead Management: far maturare i lead 19. Lead Management: far maturare i lead
19. Lead Management: far maturare i lead
 
I Mille volti della Nutrizione - Integrazione alimentare nelle patologie me...
I Mille volti della Nutrizione -   Integrazione alimentare nelle patologie me...I Mille volti della Nutrizione -   Integrazione alimentare nelle patologie me...
I Mille volti della Nutrizione - Integrazione alimentare nelle patologie me...
 
Anomali Detect 2016 - Borderless Threat Intelligence
Anomali Detect 2016 - Borderless Threat IntelligenceAnomali Detect 2016 - Borderless Threat Intelligence
Anomali Detect 2016 - Borderless Threat Intelligence
 
The Role of Design in Crowdsourcing
The Role of Design in CrowdsourcingThe Role of Design in Crowdsourcing
The Role of Design in Crowdsourcing
 
Анализ запросов российских пользователей на туры в различные страны в посиков...
Анализ запросов российских пользователей на туры в различные страны в посиков...Анализ запросов российских пользователей на туры в различные страны в посиков...
Анализ запросов российских пользователей на туры в различные страны в посиков...
 
Tanzania election poll report english version
Tanzania election poll report english versionTanzania election poll report english version
Tanzania election poll report english version
 
Press statement (english) Ipsos Tanzania september 2015
Press statement  (english)  Ipsos Tanzania september 2015Press statement  (english)  Ipsos Tanzania september 2015
Press statement (english) Ipsos Tanzania september 2015
 

Último

Future Visions: Predictions to Guide and Time Tech Innovation, Peter Udo Diehl
Future Visions: Predictions to Guide and Time Tech Innovation, Peter Udo DiehlFuture Visions: Predictions to Guide and Time Tech Innovation, Peter Udo Diehl
Future Visions: Predictions to Guide and Time Tech Innovation, Peter Udo Diehl
Peter Udo Diehl
 
Breaking Down the Flutterwave Scandal What You Need to Know.pdf
Breaking Down the Flutterwave Scandal What You Need to Know.pdfBreaking Down the Flutterwave Scandal What You Need to Know.pdf
Breaking Down the Flutterwave Scandal What You Need to Know.pdf
UK Journal
 

Último (20)

The Value of Certifying Products for FDO _ Paul at FIDO Alliance.pdf
The Value of Certifying Products for FDO _ Paul at FIDO Alliance.pdfThe Value of Certifying Products for FDO _ Paul at FIDO Alliance.pdf
The Value of Certifying Products for FDO _ Paul at FIDO Alliance.pdf
 
ECS 2024 Teams Premium - Pretty Secure
ECS 2024   Teams Premium - Pretty SecureECS 2024   Teams Premium - Pretty Secure
ECS 2024 Teams Premium - Pretty Secure
 
Future Visions: Predictions to Guide and Time Tech Innovation, Peter Udo Diehl
Future Visions: Predictions to Guide and Time Tech Innovation, Peter Udo DiehlFuture Visions: Predictions to Guide and Time Tech Innovation, Peter Udo Diehl
Future Visions: Predictions to Guide and Time Tech Innovation, Peter Udo Diehl
 
Using IESVE for Room Loads Analysis - UK & Ireland
Using IESVE for Room Loads Analysis - UK & IrelandUsing IESVE for Room Loads Analysis - UK & Ireland
Using IESVE for Room Loads Analysis - UK & Ireland
 
Simplified FDO Manufacturing Flow with TPMs _ Liam at Infineon.pdf
Simplified FDO Manufacturing Flow with TPMs _ Liam at Infineon.pdfSimplified FDO Manufacturing Flow with TPMs _ Liam at Infineon.pdf
Simplified FDO Manufacturing Flow with TPMs _ Liam at Infineon.pdf
 
Speed Wins: From Kafka to APIs in Minutes
Speed Wins: From Kafka to APIs in MinutesSpeed Wins: From Kafka to APIs in Minutes
Speed Wins: From Kafka to APIs in Minutes
 
Microsoft CSP Briefing Pre-Engagement - Questionnaire
Microsoft CSP Briefing Pre-Engagement - QuestionnaireMicrosoft CSP Briefing Pre-Engagement - Questionnaire
Microsoft CSP Briefing Pre-Engagement - Questionnaire
 
Easier, Faster, and More Powerful – Notes Document Properties Reimagined
Easier, Faster, and More Powerful – Notes Document Properties ReimaginedEasier, Faster, and More Powerful – Notes Document Properties Reimagined
Easier, Faster, and More Powerful – Notes Document Properties Reimagined
 
Breaking Down the Flutterwave Scandal What You Need to Know.pdf
Breaking Down the Flutterwave Scandal What You Need to Know.pdfBreaking Down the Flutterwave Scandal What You Need to Know.pdf
Breaking Down the Flutterwave Scandal What You Need to Know.pdf
 
IESVE for Early Stage Design and Planning
IESVE for Early Stage Design and PlanningIESVE for Early Stage Design and Planning
IESVE for Early Stage Design and Planning
 
Optimizing NoSQL Performance Through Observability
Optimizing NoSQL Performance Through ObservabilityOptimizing NoSQL Performance Through Observability
Optimizing NoSQL Performance Through Observability
 
Unpacking Value Delivery - Agile Oxford Meetup - May 2024.pptx
Unpacking Value Delivery - Agile Oxford Meetup - May 2024.pptxUnpacking Value Delivery - Agile Oxford Meetup - May 2024.pptx
Unpacking Value Delivery - Agile Oxford Meetup - May 2024.pptx
 
Extensible Python: Robustness through Addition - PyCon 2024
Extensible Python: Robustness through Addition - PyCon 2024Extensible Python: Robustness through Addition - PyCon 2024
Extensible Python: Robustness through Addition - PyCon 2024
 
Overview of Hyperledger Foundation
Overview of Hyperledger FoundationOverview of Hyperledger Foundation
Overview of Hyperledger Foundation
 
Google I/O Extended 2024 Warsaw
Google I/O Extended 2024 WarsawGoogle I/O Extended 2024 Warsaw
Google I/O Extended 2024 Warsaw
 
Oauth 2.0 Introduction and Flows with MuleSoft
Oauth 2.0 Introduction and Flows with MuleSoftOauth 2.0 Introduction and Flows with MuleSoft
Oauth 2.0 Introduction and Flows with MuleSoft
 
FDO for Camera, Sensor and Networking Device – Commercial Solutions from VinC...
FDO for Camera, Sensor and Networking Device – Commercial Solutions from VinC...FDO for Camera, Sensor and Networking Device – Commercial Solutions from VinC...
FDO for Camera, Sensor and Networking Device – Commercial Solutions from VinC...
 
Behind the Scenes From the Manager's Chair: Decoding the Secrets of Successfu...
Behind the Scenes From the Manager's Chair: Decoding the Secrets of Successfu...Behind the Scenes From the Manager's Chair: Decoding the Secrets of Successfu...
Behind the Scenes From the Manager's Chair: Decoding the Secrets of Successfu...
 
Secure Zero Touch enabled Edge compute with Dell NativeEdge via FDO _ Brad at...
Secure Zero Touch enabled Edge compute with Dell NativeEdge via FDO _ Brad at...Secure Zero Touch enabled Edge compute with Dell NativeEdge via FDO _ Brad at...
Secure Zero Touch enabled Edge compute with Dell NativeEdge via FDO _ Brad at...
 
Demystifying gRPC in .Net by John Staveley
Demystifying gRPC in .Net by John StaveleyDemystifying gRPC in .Net by John Staveley
Demystifying gRPC in .Net by John Staveley
 

Instrumenting the Crowd + Task Fingerprinting Overview