Inferring and Propagating Pivot Dependencies in Schema Transformation between Data and Metadata

To study data dependencies over the data in schema transformation between data and metadata (DM-transformation for short), we define a general dependency form, namely pivot dependency (PD for short), which specifies constraints not only on values but also on names of attribute and relation. PDs, an extension to conventional FDs, formalize constraints in context that data and metadata are treated uniformly. We first provide a complete set of inference rules of PDs analogous to Armstrong´s axioms for traditional FDs and then study the derivation of PDs during DM-transformation. Meanwhile, we discuss the lossless and nonredundant transformation and also give an approach to simplify a transformation into a non-redundant one using algebraic laws of transformation operators. The main goal of our work is to lay the theoretical foundation for DM-transformation. Our work can be used in different stages of building and accessing a database system, e.g. to detect and resolve value inconsistency in data integration involving metadata, to ensure the correctness of DM-transformation, to normalize integrated schemas, to verify the integrity of data, and to optimize queries at an integration level.