Algorithm for discovering multivalued dependencies
Men Hin Yan, Ada Wai-Chee Fu · 2001
INTRODUCTION While extracting functional dependencies has received considerable attention [6, 3, 4, 5], relatively less research effort has been put in finding multivalued dependencies. Based on the previously proposed techniques of discovering functional dependencies [3], we propose a new algorithm for finding multivalued dependencies from a large database. We assume the usual interpretation of a relation (or table) in the relational database model where no duplicate tuples are allowed. The definition of multivalued dependency is given as below: Let R be a relation schema and let X = X1 ; X2 ; :::; Xn be a subset of R , let Y = Y1 ; Y2 ; :::; Yn be a subset of R and let Z = R \\Gamma Y \\Gamma X. The multivalued dependency<