Stanford InfoLab Publication Server

Exploiting Dependencies to Enhance View Self-Maintainability

Huyn, N. (1996) Exploiting Dependencies to Enhance View Self-Maintainability. Technical Report. Stanford InfoLab. (Publication Note: Detailed version of a paper submitted for publication.)

BibTeXDublinCoreEndNoteHTML

[img]
Preview
PDF
386Kb

Abstract

View self-maintenance is the process of incrementally refreshing a materialized view using the view instance and the update to some base relation, but without examining any of the base relations, or using only a specified subset of the base relations. A new problem is created when base data use is limited in view maintenance: can the view be maintained at all? When a base relation is not available for use in maintaining a view, the next-best form of knowledge that may be available is integrity constraints the relation is known to satisfy. Constraints are often available in practice, yet very little is known about how they can be exploited fully to enhance view self-maintainability The problem becomes even harder when the SM tests are required to be in some effcient query form, e.g., SQL queries. In this paper, we focus on the problem of determining view SM in the presence of functional dependencies. We first show that SM of a conjunctive-query (CQ) view can be reduced to a problem of query containment, whose solution can be expressed as a (boolean) query on the view in safe, nonrecursive Datalog :6= . We then show that for the special class of CQ views with no repeated predicates, two useful concepts can be defined: the well-founded derivation DAG and subgoal partitioning. We then derive three simple conditions (each expressible as the union of CQ's with 6= comparisons) that each guarantee view SM under general functional dependencies. View maintenance expressions are shown to be simple CQ's. It turns out that these three conditions alone completely characterize view self-maintainability.

Item Type:Techreport (Technical Report)
Subjects:Computer Science > Query Processing
Projects:Information Integration
Related URLs:Project Homepagehttp://infolab.stanford.edu/serf/
ID Code:136
Deposited By:Import Account
Deposited On:25 Feb 2000 16:00
Last Modified:08 Dec 2008 15:39

Download statistics

Repository Staff Only: item control page