Identificação de técnicas para detecção automática de fraudes através de ontologias

Abstract

The occurrences of fraud in organizations have become quite complex and challenging, is the increase in the number of cases identified, either by increasing the level of sophistication of them. This has led to the development by public and private organizations, initiatives regarding both the prevention (measures taken to inhibit the occurrence of fraud in the first instance) as for the detection (measures taken to an event that fraud is identified as soon as possible). These initiatives involve the enactment of specific laws on the subject, developing frameworks for managing the lifecycle of fraud (with the detailed identification of all stages involved), by going to the implementation of techniques for automatic detection of fraud (based on statistical, artificial intelligence and data mining). However, in relation to policies for compliance and the frameworks to manage problems of fraud, there is a movement of consolidation and standardization of knowledge and procedures in regard to techniques for automatic detection of fraud, there is a gap in the effective relationship between the technique and mastery of the problem of fraud. Within this context, this dissertation proposes: an ontology to problems of fraud, one for automatic detection techniques of fraud, and a third ontology generated from the merging between the first two. This ontologies aim at providing the bases of knowledge to be used in applications capable of meeting the fundamental question: what or where the automatic detection techniques are more appropriate to a field of specific problem of fraud? The ontologies implemented in this work were prepared based on the methodology Uschold & King's method, the coded language Web Ontology Language (OWL), and with support from the Protégé tool version 4.0 beta. Also composed this dissertation, a proposal to extend the framework of assessment methodologies, developed by Gómez-Pérez, Fernández-López and Corcho, in order to obtain a scheme with the ability to suggest what the most appropriate methodology for building an ontology in a given context