Query rewriting using views in the presence of inclusion dependencies

  • Authors:
  • Qingyuan Bai;Jun Hong;Michael F. McTear

  • Affiliations:
  • University of Ulster at Jordanstown, Antrim, UK;University of Ulster at Jordanstown, Antrim, UK;University of Ulster at Jordanstown, Antrim, UK

  • Venue:
  • WIDM '03 Proceedings of the 5th ACM international workshop on Web information and data management
  • Year:
  • 2003

Quantified Score

Hi-index 0.00

Visualization

Abstract

Query rewriting using views is an essential issue in data integration. A number of algorithms, e.g., the bucket algorithm, the inverse rules algorithm, the SVB algorithm and the MiniCon algorithm, have been proposed to address this issue. These algorithms can be divided into two categories: bucket-based algorithms and inverse rule-based algorithms. Some inverse rule-based algorithms have considered the problem of query rewriting in the presence of inclusion dependencies. However, there has been no bucket-base algorithm so far for the problem. All the previous bucket-based algorithms may miss query rewritings in the presence of inclusion dependencies. In this paper, we extend the MiniCon algorithm to the presence of inclusion dependencies. In the MiniCon algorithm, a view can be used in a non-redundant rewriting of a query only if at least one subgoal in the query is covered by a subgoal in the view. In the presence of inclusion dependencies, when no subgoal in a view directly covers the query subgoal we can apply the chase procedure and rule to the subgoals of the query or view that contains the chase reachable subgoals to get a revised query or view. The condition required by the MiniCon algorithm is then satisfied. We can therefore avoid the problem of missing rewritings with the previous bucket-based algorithms. We prove that our extended algorithm can find the maximally-contained rewriting of a conjunctive query using a set of conjunctive views in the presence of inclusion dependencies. Our extension of the MiniCon algorithm does not involve a significant increase in computational complexity and our new algorithm remains scalable.