Finding Where to Apply Object-Relational Database Schema Refactorings: an Ontology-Guided Approach

Less complex object-relational (OR) database schemas are more understandable, changeable, maintainable, and reusable. This paper addresses the use of OR metrics to (i) detect complex fragments of OR database schemas, which are candidates to perform refactorings, and (ii) to assess if the application of the selected refactorings has indeed reduced the schema complexity. The schema metrics are formally defined with OCL, upon an ontology of the SQL:2003 standard, expressed as a UML class diagram. This formalization allows automating metrics collection. Although our approach does not point out how to select the most adequate refactoring transformation, it helps designers and developers pinpoint the spots in the database schema where refactoring is likely to produce a quality improvement. An illustrating example is included