Influences of functional dependencies on bucket-based rewriting algorithms

Abstract

There are several algorithms for the problem of query rewriting using views, including the bucket-based algorithms and the inverse rule-based algorithms. The inverse rule-based algorithms have considered the problem in the presence of functional dependencies. However, there are few papers that consider the influences of functional dependencies from the point of view of the bucket-based algorithms. In this paper, we discuss this issue in the bucket-based framework. We mainly consider how to use conditions weaker than the ones in the previous algorithms when forming a set of buckets. The novelty of our approach is that we can enforce additional constraints on the non-exported attributes in a view through functional dependencies. As a result, more valid buckets can be formed. We can then generate additional rewritings which cannot be generated by the previous algorithms. We prove that our algorithm is sound and complete, i.e., the union of the obtained rewriting is a maximally-contained rewriting relative to functional dependencies under certain assumptions. In addition, we also discuss how to construct a recursive rewriting by making use of functional dependencies. © Springer-Verlag Berlin Heidelberg 2005.