Issue No. 04 - December (1991 vol. 3)
DOI Bookmark: http://doi.ieeecomputersociety.org/10.1109/69.109108
<p>The authors investigate the inference problems due to functional dependencies (FD) and multivalued dependencies (MVD) in a multilevel relational database (MDB) with attribute and record classification schemes, respectively. The set of functional dependencies to be taken into account in order to prevent FD-compromises is determined. It is proven that incurring minimum information loss to prevent compromises is an NP-complete problem. An exact algorithm to adjust the attribute levels so that no compromise due to functional dependencies occurs is given. Some necessary and sufficient conditions for MVD-compromises are presented. The set of MVDs to be taken into account for controlling inferences is determined. An algorithm to prevent MVD-compromises in a relation with conflict-free MVDs is given.</p>
FD inferences; attribute classification; MVD inferences; functional dependencies; multivalued dependencies; multilevel relational database; record classification; FD-compromises; minimum information loss; NP-complete problem; exact algorithm; necessary and sufficient conditions; MVD-compromises; conflict-free MVDs; computational complexity; database theory; inference mechanisms; relational databases
G. Ozsoyoglu and T. Su, "Controlling FD and MVD Inferences in Multilevel Relational Database Systems," in IEEE Transactions on Knowledge & Data Engineering, vol. 3, no. , pp. 474-485, 1991.