A STUDY ON FRAUDSTER IDENTIFICATION IN ELECTRONIC MARKETS THROUGH HUMAN COMPUTATION / UM ESTUDO DE IDENTIFICAÇÃO DE FRAUDADORES EM MERCADOS ELETRÔNICOS ATRAVÉS DA COMPUTAÇÃO HUMANA

AUTOR(ES)
DATA DE PUBLICAÇÃO

2008

RESUMO

Fraudulent behavior is an increasing problem for electronic markets, in particular for online auction sites, causing several types of loss. Fraud loss reduction measures generally have as an undesirable by-product the harassment and even exclusion of bona fide users, creating a difficult trade- off between losses with fraudsters and losses due to excessive constraints on market participants. The objective of this thesis is to show the viability of a novel approach to fraud loss reduction in online auction sites, the catch the thief game. This approach takes explicitly into account the aforementioned trade-off and is based on the paradigm of human computation, where people do computational tasks for fun or profit. The methodology used was an exploratory research on fraudulent activity in a real electronic market, a pilot test of fraudster detection by human agents, and the development and simulation of the proposed game s core element, the fraudster identification mechanism. The exploratory research presents a profile of non-delivery fraud in the biggest Brazilian online auction site, showing it as real, recurring and measurable problem; the pilot test displays positive evidence that unspecialized human agents can indeed distinguish fraudulent sellers from normal ones by a significant margin; the simulation supports the usefulness of the proposed mechanism for fraud loss reduction. The results obtained confirm catch the thief game as a viable approach to reduce fraud loss in electronic markets.

ASSUNTO(S)

frauds comercio eletronico e-commerce computacao humana fraudes human computing

ACESSO AO ARTIGO

Documentos Relacionados