Application of Data Mining in Tax Processes Improvement: A Literature Review and Classification

Data mining is an effective tool to improve and enhance the efficiency and effectiveness of tax processes by extracting beneficial knowledge and insight from tax data. The purpose of this paper is to study the status of research pieces in the field, classify them, identify the research gaps and provide a roadmap for researchers and practitioners through a systematic literature review.

Due to the importance of the subject, this study focuses on studies conducted in the field of data mining and taxation from 2000 to 2021. It investigates their processes and practical domains. The reviewed studies were categorized based on the proposed framework and various analyses were presented in terms of processes, practical domains, and data mining techniques. Furthermore, the distribution of papers according to the year of their publication and also regarding the journal in which they were published were presented. Tax processes were divided into four groups i.e. submission, examination, collection, and taxpayer services. The defined practical domains included tax payment, tax refund, shell corporation identification, identification of non-filer taxpayers, risk-based tax audit selection, tax debt management, and tax comments analysis. The classification framework for data mining techniques in this study was defined as clustering, association analysis, classification and prediction, regression, time series, anomaly detection, and visualization.

According to the obtained findings, most of the reviewed studies were assigned to the inspection process, 94 percent of which worked on the practical domain of “risk-based tax audit selection”. The most popular and widely used technique was "classification and prediction", while the three algorithms including neural network, decision tree, and support vector machine were widely used, compared to other algorithms.

Currently, tax administrations have huge databases and traditional methods and tools cannot analyze them due to the limited resources of organizations as well as the large amounts of available data. Data mining can have an effective performance on various tax processes and can be effective in making decisions and adopting appropriate approaches. There is good potential for the application of data mining techniques in all of the proposed practical domains. In the submission and collection processes, more research needs to be done. Some approaches including reinforcement learning, deep learning, graph analysis, and big data analytics are recommended for future research. Proposing practical frameworks for using data mining techniques in tax systems and tax administrations is also recommended. To the best of the author's knowledge, no study has been conducted to investigate the issue, while there is a definite need in this regard. Besides, one of the important issues, which needs to be addressed as the main gap in this field, is integrating the internal and external sources of data, which can improve effectiveness.