Agent based fraud detection and reporting in public e-procurement

Abstract

Procurement fraud remains an endemic in most modern economies. E-Procurement fraud may manifest in different ways that can include collusion by parties involved in procurement as well as falsification of documents. A procurement officer might be induced through bribery to favor a particular supplier. For protection against procurement fraud, organizations have tried to implement some control measures, hoping such measures would discourage the fraud that is directed on institutions. Complex fraud does not revolve around the breaching of these controls, but bypassing them. In this research we set out to design and implement an e-procurement fraud detection tool for public entities using multi agent technologies. This was informed by contributions from various government employees who were interviewed, literature review and publications that indicate the presence of fraud in public offices attributable to procurement processes. A prototype of an e-procurement system is developed with the complete procure-topay functionality. This provides the environment for the agent based fraud detection tool to be implemented on. Fraud is then detected using rule set to determine suspicious activities and transactions in the e-procurement system. The agent based e-procurement fraud detection tool is able to detect and report fraud in situations where inflation of unit cost of items at requisition level and further upward adjustments are done while raising purchase orders. Upward adjustment of quantities on purchase orders after requisition approval is also picked as fraud by the agent detection tool. This is a scenario that requires approvals from approvers who may be compromised or fail to take note of the discrepancies. The proceeds from such fraud may be paid to the participants in the procurement chain as kickbacks (bribes).