On simplification of schema mappings

  • Authors:

  • Diego Calvanese;Giuseppe De Giacomo;Maurizio Lenzerini;Moshe Y. Vardi

  • Affiliations:

  • KRDB Research Centre for Knowledge and Data, Free University of Bozen-Bolzano, Piazza Domenicani 3, I-39100 Bolzano, Italy;Dipartimento di Informatica e Sistemistica, Sapienza Universití di Roma, Via Ariosto 25, I-00185 Roma, Italy;Dipartimento di Informatica e Sistemistica, Sapienza Universití di Roma, Via Ariosto 25, I-00185 Roma, Italy;Dep. of Computer Science, Rice University, P.O. Box 1892, Houston, TX 77251-1892, USA

  • Venue:
  • Journal of Computer and System Sciences
  • Year:
  • 2013

Quantified Score

Hi-index 0.00

Visualization

Abstract

A schema mapping is a formal specification of the relationship holding between the databases conforming to two given schemas, called source and target, respectively. While in the general case a schema mapping is specified in terms of assertions relating two queries in some given language, various simplified forms of mappings, in particular lav and gav, have been considered, based on desirable properties that these forms enjoy. Recent works propose methods for transforming schema mappings to logically equivalent ones of a simplified form. In many cases, this transformation is impossible, and one might be interested in finding simplifications based on a weaker notion, namely logical implication, rather than equivalence. More precisely, given a schema mapping M, find a simplified (lav, or gav) schema mapping M^' such that M^' logically implies M. In this paper we formally introduce this problem, and study it in a variety of cases, providing techniques and complexity bounds. The various cases we consider depend on three parameters: the simplified form to achieve (lav, or gav), the type of schema mapping considered (sound, or exact), and the query language used in the schema mapping specification (conjunctive queries and variants over relational databases, or regular path queries and variants over graph databases). Notably, this is the first work on comparing schema mappings for graph databases.