چكيده به لاتين
Abstract
Nowadays, tax auditors face different challenges like accurate tax identification and collection of various businesses which do tax evasion. Against with them, tax auditors have limited resources and choose traditional tax integrity verification approaches. These approaches are almost cost and time consuming, but modern approaches are avilable using computer techniques to detect tax evasion dubiouses in a cost effective fast way.
In this study, different data mining and machine learning techniqueshave been used and proposed as an effective approach to challenge tax evasion and perform fraud detection for tax administration. Data mining models, algorithms and validation approaches are widely described and at last, a data mining framework has been introduced to reduce tax evasion by detecting tax evasion dubiouses. This general framework is based on fraud detection and tax evasion reduction processes, so we tried to present a general framework to set various types of tax evasion reduction objectives. At last, we provided information transform into the knowledge as an output. This framework also applied on Iran’s tax administration data as a practical and valuable data source for tax mining to reduce tax evasion.
Keywords: data mining – tax evasion- fraud detection - framework
