“…However, many researchers believed that manual case selection and whistle-blowingbased selection are time-consuming and tedious, while data mining techniques used by tax administrations to detect tax fraud are considered to be the most promising approaches [7]. Mechanisms, such as neural networks, decision trees [8], logistic regression, SOM (Selforganizing map), K-means, support vector machines, visualization techniques, Bayesian networks, rough set [3], K-nearest neighbor, association rules [24], fuzzy rules, Markov chains, time series, regression and simulations [2], have been used to check tax evasion [23], [12]. For example, Wu et al [23] used a data mining technique and developed a screening framework to filter possible noncompliant value-added tax (VAT) reports that may be subject to further auditing.…”