Collusion and Corruption Risk Analysis Using Naïve Bayes Classifiers

Fighting corruption connected with public procurement and governmental agencies requires a strong and effective audit function. The scale and the complexity of the roles to be considered can prevent the use of most audit methods and technologies so successful on the corporate world. The aim of this chapter is to propose a data mining method, based on naive Bayes classifiers, to support a generic risk assessment process for audit planning. The method can sort auditable units by total risk score, fostering dedicated audit coverage to high-risk areas. Audit organizations can transition from a reactive response to a proactive approach to identify and correct issues that may be indicative of fraud, waste or abuse. Extensive databases containing records from government operations can be combined to auditors knowledge, fraud profiles, impact factors or any other relevant metric in order to rank an audit universe.