Using a knowledge learning framework to predict errors in database design

• Determines the relationship of designer expertise on ER modeling errors.
• Successfully predicts high/medium frequency errors caused by incorrect representation.
• Provides a methodology to measure task complexity and expertise with Bloom’s taxonomy.
• Provides guidelines for effectively structuring teaching modules for ER modeling.

Conceptual data modeling is a critical but difficult part of database development. Little research has attempted to find the underlying causes of the cognitive challenges or errors made during this stage. This paper describes a Modeling Expertise Framework (MEF) that uses modeler expertise to predict errors based on the revised Bloom's taxonomy (RBT). The utility of RBT is in providing a classification of cognitive processes that can be applied to knowledge activities such as conceptual modeling. We employ the MEF to map conceptual modeling tasks to different levels of cognitive complexity and classify current modeler expertise levels. An experimental exercise confirms our predictions of errors. Our work provides an understanding into why novices can handle entity classes and identifying binary relationships with some ease, but find other components like ternary relationships difficult. We discuss implications for data modeling training at a novice and intermediate level, which can be extended to other areas of Information Systems education and training.