EPFL doctorate Award 2009 - Jurca Radu

"For developing a methodology for obtaining truthful information in a wide variety of practical settings."

Truthful reputation mechanisms for online systems.

Reputation mechanisms collect, distill and disseminate the information available in a market about the behavior and the characteristics of market participants. They provide better information to decision makers, increasing their chances of making the right choices, and helping them avoid deception. The recent growth of internet-based communities and economies promoted reputation mechanisms as an effective means for establishing trust in online environments.

The economical value of online reputation raises questions regarding the trustworthiness of the mechanism itself. Existing systems were conceived with the assumption that users will share feedback honestly. However, recent evidence suggests that some users strategically manipulate their feedback to game the mechanism.

This thesis describes ways of making online reputation mechanisms more trustworthy by providing rational incentives for reporting the truth. It studies different kinds of reputation mechanisms, and different reward schemes that promote honest reporting. Problems related to collusion (i.e., several agents coordinate their strategies to manipulate reputation information) and robustness are also investigated.
Moreover, the thesis describes a novel application of incentive compatible reputation mechanisms to the area of quality of service monitoring, and investigates factors that motivate and bias human users when reporting feedback in existing review forums.