AI-Based Data Quality Assurance for Business Intelligence and Decision Support Systems

BI and DSS are valuable tools that help businesses make better decisions during the present information-oriented economy. However, these systems are highly dependent on the kind of data they receive and process. This means that one wrong piece of information collected creates wrong information, leading to wrong decisions and losses. Therefore, this paper reviews AI data quality assurance relevant to BI and DSS. The paper reviews data quality criteria, including accuracy, completeness, consistency, timeliness, and validity. Using such works as machine learning, deep learning, natural language processing, and knowledge graphs in data quality assurance is thoroughly studied. A new approach for data pre-processing has been developed, comprising anomaly detection, data cleansing, intelligent imputation, and semantic reconciliation to enhance data quality. Thus, we decided to conduct a literature review aimed at assessing the existing methodologies and identifying the deficits that require the use of AI. In order to address the data concerns in the implementation of training algorithms, the proposed methodology focuses on using supervised and unsupervised learning to identify the problems of data and rectify the same simultaneously. Some studies, based on experiments carried out for a BI system on a large-scale data set of a financial institution, showed better data quality and accuracy of decision-making. Moreover, we provide recommendations, drawbacks, and prospects for the application of the augmentation of AI in data quality assurance