IBM®
Skip to main content
    Country/region [change]    Terms of use
 
 
 
    Home    Products    Services & solutions    Support & downloads    My account    

IBM Technical Journals

Special report: Celebrating 50 years of the IBM Journals
All topics > Fundamental Science and Mathematics >

Functional dependencies in a relational database and propositional logic

Award plaque by R. Fagin

An equivalence is shown between functional dependency statements of a relational database, where “ → ” has the meaning of “determines,” and implicational statements of propositional logic, where “ ⇒ ” has the meaning of “implies.” Specifically, it is shown that a dependency statement is a consequence of a set of dependency statements iff the corresponding implicational statement is a consequence of the corresponding set of implicational statements. The database designer can take advantage of this equivalence to reduce problems of interest to him to simpler problems in propositional logic. A detailed algorithm is presented for such an application. Two proofs of the equivalence are presented: a “syntactic” proof and a “semantic” proof. The syntactic proof proceeds in several steps. It is shown that 1) Armstrong's Dependency Axioms are complete for dependency statements in the usual logical sense that they are strong enough to prove every consequence, and that 2) Armstrong's Axioms are also complete for implicational statements in propositional logic. The equivalence then follows from 1) and 2). The other proof proceeds by considering appropriate semantic interpretations for the propositional variables. The Delobel–Casey Relational Database Decomposition Theorems, which heretofore have seemed somewhat fortuitous, are immediate and natural corollaries of the equivalence. Furthermore, a counterexample is demonstrated, which shows that what seems to be a mild extension of the equivalence fails.

Originally published:

IBM Journal of Research and Development, Volume 21, Issue 6, pp. 534-544 (1977).

Significance:

Theoretical database studies form a fundamental framework for the development of sound database management systems, which have provided increasingly sophisticated capabilities for data handling. Studies of database design challenges require a precise formalization so that detailed analyses and comparisons can be made. Such analyses facilitate the design of superior database management systems in terms of performance, reliability, new ways of handling query evaluations and incomplete information, and the specification of data models and data dependencies.

This highly cited paper introduces and describes the strong relationship between mathematical logic and database function. Fagin and his colleagues subsequently broadened the IBM Journal paper findings in a 1981 paper published in the Journal of the ACM (“An equivalence between relational database dependencies and a fragment of propositional logic,” Vol. 28, No. 3, pp. 435-453). Fagin's IBM Journal paper was the first to introduce concepts and notations from mathematical logic, such as logical implication, for the study of databases. Mathematical logic has subsequently been shown to provide a fundamental tool in the study of databases, and many papers have described how to best take advantage of this connection between logic and database design.

Comments:

Related paper: An equivalence between relational database dependencies and a fragment of propositional logic (JACM 1981) by Y. Sagiv, C. Delobel, D. S. Parker, and R. Fagin


    About IBMPrivacyContact