Mineração de dados para detecção de fraudes em transações eletrônicas

• Ano de defesa: 2012
• Autor(a) principal: José Felipe Júnior
• Tipo de documento: Dissertação
• Tipo de acesso: Acesso aberto
• Idioma: por
• Instituição de defesa: Universidade Federal de Minas Gerais; UFMG
• Palavras-chave em Português: Comércio Eletrônico; Aplicação Web; Mineração de Dados; KDD; Detecção de Fraudes; Comércio eletrônico (Computação); Computação; Mineração de dados (Computação)
• Link de acesso: http://hdl.handle.net/1843/ESBF-8SYPK3

With Internet popularity, the number of people who use it to make financial transactions increase every year. This happen due to the facilities promoted by the Web to make purchases and payments at any time and anywhere. However, this popularity has been attracted criminals attention resulting in a significantly increased the number of fraud cases in this scenario. The worldwide financial losses reach billions of dollars per year, which shows the need for a study of fraud in the light of this new context. This work use the knowledge discovery process for fraud detection in online payments. More than that, is a comprehensive survey conducted in the fraud detection area where we worrying about issues ranging from the database, passing by the evaluation of the most promising techniques, to issues related to the financial return to the techniques. As a way of evaluating the proposed approach, we defined a concept of economic efficiency and applied to a real dataset from one of the largest Brazilian companies in electronic payment services. The results show a good performance in detecting fraud, with gains in excess of the current situation of the company.