Modeling of Crowdsourcing Platforms and Granularity of Work Organization in Future Internet

Transcription

1 Institute of Computer Science Chair of Communication Networks Prof. Dr.-Ing. P. Tran-Gia Modeling of Crowdsourcing Platforms and Granularity of Work Organization in Future Internet Tobias Hoßfeld, Matthias Hirth, Phuoc Tran-Gia www3.informatik.uni-wuerzburg.de

2 Crowdsourcing in the News ReadWriteWeb.com Wall Street Journal globaleconomicanalysis.blogspot.com Venture Beat New York Times 2 2

3 Outline Introduction to crowdsourcing Modeling growth of crowdsourcing platforms Activity model for crowdsourcing platform users Conclusion 3 3

4 Evolution of Work Organization Employer (E) Mediating Platforms Contractor/ Worker (W) Project E negotiates with W E chooses W Outsourcing firm Tasks E negotiates with W E chooses W Outsourcing firm Subtasks E chooses W through platform Outtasking Platform W gets paid through platform Freelancer Microtasks Crowdsourcing Platform Control Engine Large crowd 4 4

5 Evolution of Work Organization Differences to traditional work organization Very fine granularity of tasks Employer does not select designated worker Mediator between worker and employer Workers choose tasks to work on Questions Employer: Will my task finish in time? Completion time of tasks Platform operator: Will my platform be successful? Platform growth Activity of platform users 5 5

6 Modeling growth of crowdsourcing platforms 6

7 Growth Models for Microworkers.com Measurement data User development of Microworkers.com Growth models Bio-inspired Exponential growth model Logistic growth model Artificial Square growth model 7 7

8 Exponential Growth Model Large deviations between model and measurements Exponential growth model Measurement data 8 8

9 Logistic Growth Model Small over estimation Logistic growth model Measurement data 9 9

10 Square Growth Models Square growth model Measurement data 10 10

11 Growth Estimation Exponential growth model: Exponential growth model Square growth model One million users in 2013 Logistic growth model: End of growth: 2012 Number of users: Logistic growth model About 130,000 Square growth model: One million users in

12 Activity model for crowdsourcing platform users 12

13 Modeling of Crowdsourcing Dynamics Crowdsourcing platform Non - registered users N Active users A Inactive users I System description: 13 13

14 Modeling of Crowdsourcing Dynamics Crowdsourcing platform Non - registered users N Active users A Inactive users I User interaction models: Local User Decision (LUD) User only influenced by personal opinion Globally Influenced Dynamics (GID) Users are influence by global opinion 14 14

15 Platform Dynamics using the LUD Model Nonregistered users Active users Steady state Inactive users 15 15

16 Conclusion 16

17 Conclusion Crowdsourcing platforms are A consequence of a new granularity of work organization in the Internet Growth of crowdsourcing platforms Crowdsourcing platforms could become future traffic hotspots Activity of the users on the platforms Estimation of percentage of active users in the steady state Completion times of tasks Completion times are hard to predict 17 17

18 Thank you for your attention! 18 18

19 19 19

20 Growth of Facebook In 2009 about 350 million users: Most visited web page form business PCs in 2009 In 2010 about 600 million users: Most used web page in the USA in