Bio-Inspired Agent-Based Architecture for Fraud Detection
Javier Cabezas, Alberto Fernández-Isabel, Rubén R. Fernández, César González-Fernández, Adrián Alonso Barriuso, Isaac Martín de Diego · 2020
The information accessible on the Internet has been drastically increased over the last years. This information is not always produced by trusted sources. This fact leads to the emergence of fraudulent websites containing unappropriated information or having malicious intentions. The analysis of this large amount of information to detect possible fraud situations tends to be a very demanding task for human experts. Thus, it becomes a key issue to automatise these operations. This paper presents a Multi-Agent System (MAS) model and its implementation focused on automatically detecting fraudulent websites. The INGENIAS methodology and the MESA framework have been selected for this purpose. The system consists of a bio-inspired agent-based architecture based on insect colonies. Several basic agents carry out simple and distributed operations, while there is just one agent which aggregates the individual outcomes to obtain the final result. Some websites have been selected to illustrate the viability of the proposal.