Artificial intelligence (AI) systems, such as chatbots or autonomous systems, are only as good as the data they are trained on, following the well-known “garbage-in, garbage-out” principle. Similarly, data quality directly impacts human decision-making across diferent domains, from healthcare to manufacturing. In the past, decision makers were still able to manually assess and interpret the quality of data at hand. However, with recent advances in digitalization and the deployment of AI systems in practice, the amount of data being collected, stored, and consequently used for automated decisionmaking, exceeds the capabilities of humans to process it. Hence, the need for automated data quality assessment and improvement methods has developed.
Acknowledgments This invited talk is based on research together with Divya Bhadauria, Carolina Cortes, Lorena Etcheverry, context of the Metis project: www.metisdq.org. Germany
CEUR