A Generalization of the Inference Rules for Join Dependencies in Databases

Abstract

In this paper a generalisation of the inference rules of the join dependencies that are used in the design of database schemas that meets the requirements of the fifth normal form is considered. In the previous works devoted to this problem, attempts to construct systems of the axioms of such dependencies based on inference rules are made. However, while the justification for the consistency (soundness) of the obtained axioms does not cause difficulties, the proof of completeness in general has not been satisfactorily resolved. First of all, this is due to the limitations of the inference rules themselves. This work focuses on two original axiom systems presented in the works of Sciore and Malvestuto. For the inclusion dependencies a system of rules that generalises existing systems and has fewer restrictions has been obtained. The paper presents a proof of the derivability of known systems of axioms from the presented inference rules. In addition, evidence of the consistency (soundness) of these rules is provided. The question of the completeness of the formal system based on the presented rules did not find a positive solution. In conclusion, the theoretical and practical significance of the inference rules for the join dependencies is noted.

Keywords

• relational databases
• join dependencies
• inference rules