Multi-Agent System For Fraud Detection In The Public Administration

This paper presents the initial modeling of a Multi-agent System (MAS) using the JaCaMo Platform applied to fraud detection. The MAS integrates a fraud detection system developed under the Céos Project , which identifies new data in databases and verifies information through queries and cross-referencing, applying statistical and artificial intelligence techniques. In particular, agents in the MAS simulate prosecutors' actions, investigating potential fraud through evidence-based trails, flagging only cases with significant evidence and reducing the workload of human prosecutors in the investigation process.