Author(s)

H. Hinrichs & T. Wilkens

Abstract

Data quality is one of the most challenging problems today's organisations have to cope with. Knowledge discovery in databases and data mining seem to be promis- ing concepts with regard to error detection and correction in databases. In this paper, we propose a process model for mining-based data cleansing, so-called data auditing, which we derived from the "classical" KDD model. Furthermore, we present a software system for data auditing that implements this process model. The auditing system makes extensive use of metadata, providing a flexible adapt- ability to various application domains. Our metadata model is based on an es- tablished standard to ensure system interoperability. The prototype of our data auditing system has been evaluated by means of cancer registry data. 1 Introduction The increasing popularity o

Keywords