Coding Theory Motivated by Relational Databases

Abstract

In the present paper results on minimal Armstrong instances of certain integrity constraints in relational databases are surveyed that lead to coding theory type problems. First, branching dependencies are studied. Finding minimal Armstrong instances for some collections of these integrity constraints lead to a new metric space. Error correcting codes in that space have been investigated on their own right since then.In the second part Armstrong instances of functional dependencies are investigated when the size of the domain of each attribute is bounded by a constant q. These come up naturally in real life databases, as well as in the study of higher order data model. These instances can be directly considered as q-ary codes, if tuples are taken as codewords.KeywordsRelational DatabaseFunctional DependencyIntegrity ConstraintCode TheoryUnordered PairThese keywords were added by machine and not by the authors. This process is experimental and the keywords may be updated as the learning algorithm improves.