Towards measuring test data quality

  • Authors:
  • Johannes Held;Richard Lenz

  • Affiliations:
  • University of Erlangen and Nuremberg, Erlangen;University of Erlangen and Nuremberg, Erlangen

  • Venue:
  • Proceedings of the 2012 Joint EDBT/ICDT Workshops
  • Year:
  • 2012

Quantified Score

Hi-index 0.00

Visualization

Abstract

In order to enable proper system and integration testing, it is often necessary to have huge test data inventories, reflecting the heterogeneous live system. Although the maintenance of large data stores can be guided by advice obtained from data quality evaluations, this technique can be only partly applied to test data inventories. Assessing test data quality is difficult, as the well-known data quality dimensions are not applicable in an easy fashion. For example, an otherwise good value of 100% for correctness would not allow to store erroneous test data items. The need for data quality dimensions dedicated to assessing test data quality can't be satisfied by well-known data quality dimensions. In this paper, we present our thesis approach to identify and validate new quality dimensions applicable for test data quality and develop quantification methods. We propose proximity to reality and degree of coverage as two new test data quality dimension and sketch quantification approach to measures, specifically suited for test data.