Improved Ghost Worker Fraud Detection System Using Artificial Neural Network

Abstract

Payroll fraud has been identified as one of the major problems in Nigeria’s civil service sector and it is considered as a huge drain to the economy. Over 40% of the government total recurrent expenditure is channeled to personnel cost, and ghost workers milks a large portion of this recurrent expenses. The primary aim of this work is to design and implement an improved fraud detection system for ghost workers using artificial neural network. The methodology adopted was a forensic analysis using an artificial neural network model and measuring the rules whose weight of computation to fire an artificial neural network model and further analysis and design using software development life cycle for payroll design architecture was used. Microsoft Visual Studio 2010 tools for web application which contains C# programming language, HTML, CSS, SQL server and JavaScript was used to implement the application development. The results showed web pages with a dataset of transactions to check for fraud among civil service employees and classification of the staff payroll transaction data by the trained model which detects whether a staff is a ghost worker or not. However, this forensic model application will help all department of the civil service commission to reduce the rate of fraud that is rampant in almost all the ministries of the federal republic of Nigeria and to sustain national development.