Truthful opinions from the crowds
Radu Jurca, Boi V. Faltings · ACM SIGecom Exchanges · 2008
An increasing number of applications of artificial intelligence extract knowledge from large groups of agents, also termed the wisdom of the crowds. One example are online feedback forums (also known as reputation mechanisms) for obtaining information about the products or services. The testimonies of previous buyers disclose hidden product attributes such as quality, reliability, ease of use, etc., that can only be observed after the purchase. This previously unavailable information allows the buyers to make more efficient decisions, and eliminates some of the problems that would otherwise lead to the collapse of online markets [Akerlof 1970]. Recent studies, however, raise important questions regarding the ability of existing reputation mechanisms to reflect the real quality of a product. First, the absence of clear incentives drives only some of the users to voice their opinions. For example, most Amazon ratings for a book or CD are either very good, or very bad, while controlled experiments on the same items reveal normally distributed opinions [Hu et al. 2006]. Second, some users intentionally lie to distort the public reputation in their favor. Fake reviews can be seen on Amazon [Harmon 2004],