Investigation of Application Specific Metrics to Data Quality Assessment

摘要

Databases have risen to be one of the most important corporate assets, but usually their data quality is poor or even not manageable at all. Several metrics of data quality have been designed and implemented to monitor a database of an information system. The primary goal of data quality metrics design was to provide the managers of information centres the tools for monitoring of their databases and for alerting that the amount of errors crossed a given threshold value and it is necessary to undertake activities aimed at data cleansing. The metrics should also be useful for the software producers to enable them to improve their applications. The proposed metrics have been evaluated using databases of several cadastral information systems. The investigation of data quality changes during the period of last three years is presented in the paper. Several metrics applying to domain and referential defects have been used. The study has revealed that simple metrics based on the number of defects detected in a database are not sufficient. The metrics calculating the cost of defect removal have been proposed and evaluated. The experiments covered also metrics specific for cadastral data.