Managing Quality of Crowdsourced Data
Event details
| Date | 24.08.2016 |
| Hour | 10:00 › 12:00 |
| Speaker | Naman Goel |
| Location | |
| Category | Conferences - Seminars |
EDIC Candidacy Exam
Exam President: Prof. Marcel Salathé
Thesis Director: Prof. Boi Faltings
Co-examiner: Prof. Daniel Gatica-Perez
Background papers:
Eliciting Informative Feedback: The Peer-Prediction Method, by N. Miller, P. Resnick, R. Zeckhauser
Truth Discovery with Multiple Conflicting Information Providers on the Web, by X. Yin, J. Han, P. Yu
Incentive Schemes for Participatory Sensing, by G. Radanovic, B. Faltings.
Abstract
Collecting data from people has become easier with web based technologies. However, ensuring the quality of this data is difficult. The problem isn’t limited to data elicited on crowdsourcing platforms. Web is dominated by user-generated content. This content is copied/shared/endorsed by various sources. Presence of untrustworthy sources results in low quality data. We consider one dimension of the solution space which deals with incentivizing people to report correct data. This includes designing payment mechanisms with no access to truth. A specific application scenario is participatory sensing, where a spatially distributed phenomenon is reported by the crowd and the true observation is not available to center for verification of reports. Finally, we discuss some of our preliminary results and research proposal.
Exam President: Prof. Marcel Salathé
Thesis Director: Prof. Boi Faltings
Co-examiner: Prof. Daniel Gatica-Perez
Background papers:
Eliciting Informative Feedback: The Peer-Prediction Method, by N. Miller, P. Resnick, R. Zeckhauser
Truth Discovery with Multiple Conflicting Information Providers on the Web, by X. Yin, J. Han, P. Yu
Incentive Schemes for Participatory Sensing, by G. Radanovic, B. Faltings.
Abstract
Collecting data from people has become easier with web based technologies. However, ensuring the quality of this data is difficult. The problem isn’t limited to data elicited on crowdsourcing platforms. Web is dominated by user-generated content. This content is copied/shared/endorsed by various sources. Presence of untrustworthy sources results in low quality data. We consider one dimension of the solution space which deals with incentivizing people to report correct data. This includes designing payment mechanisms with no access to truth. A specific application scenario is participatory sensing, where a spatially distributed phenomenon is reported by the crowd and the true observation is not available to center for verification of reports. Finally, we discuss some of our preliminary results and research proposal.
Practical information
- General public
- Free
Contact
- Cecilia Chapuis EDIC