A rule-based approach to analyzing database schema objects with datalog

Abstract

Database schema elements such as tables, views, triggers and functions are typically defined with many interrelations. In order to support database users in understanding a given schema, a rule-based approach for analyzing the respective dependencies is proposed using Datalog expressions. We show that many interesting properties of schema elements can be systematically determined this way. The expressiveness of the proposed analysis is exemplarily shown with the problem of computing induced functional dependencies for derived relations. The propagation of functional dependencies plays an important role in data integration and query optimization but represents an undecidable problem in general. And yet, our rule-based analysis covers all relational operators as well as linear recursive expressions in a systematic way showing the depth of analysis possible by our proposal. The analysis of functional dependencies is well-integrated in a uniform approach to analyzing dependencies between schema elements in general.