Using data samples in validating data models

Abstract

This paper presents a method for independent data model validation. The proposed adaptation of the Sharp's method is more effective than normalisation alone and easy to learn. The paper argues that regardless of modelling notation used, both conceptual and logical data models can be validated. Furthermore, the paper presents examples covering cases of normal forms violations. Among the lessons learned is recognition of patterns that can help avoid certain modelling errors. We also report initial evidence of the effectiveness of the proposed method.