Date of Publication :29th March 2018

Abstract: Tax evasion is the illegal underpayment or non-payment of tax to the government. Detecting these illegal tax evasion activities is an important as well as challenging issue for the tax administration system of any country and the biggest challenge is the increasing volume of tax data. In this paper, we propose a system that can help to characterize and detect the probable tax evaders using the information in their tax payment through big data analysis techniques. After pre-processing the available tax data, K-mean clustering algorithm is applied to form clusters of taxpayers with similar behaviour. Then decision trees are used to classify each cluster into tax payers with and without fraud and patterns in their associated behaviour are detected. Using these characteristics and patterns, the artificial neural network is trained and potential fraudsters could be detected based on the information available. This system will help detect tax fraudsters and enhance knowledge on their patterns of fraud which will be helpful in the prevention of fraud.

Reference :