Algorithms for Adapting Materialised Views in Data Warehouses

In this paper we consider the problem of materialised view adaptation in data warehouses. Materialised views are important in data warehousing where they are used to speed up query processing on large amounts of data. User requirements change over time, which may change the de nitions of views dynamically. For such situations, the question arises whether the materialised views should be recomputed from scratch for every change in the de nition or they should be obtained by adapting old materialised views. Changes to a view de nition may be expensive, if the view is recomputed from scratch. Therefore, it is worthwhile to examine ways of performing changes to the materialised view without recomputing the entire view which has undergone a change in de nition. We present adaptation algorithms for adapting views when the changes are made in each SELECT, FROM, and WHERE clause. The main idea of our algorithms is to augment the schemas of base relations by adding `joincount' attributes to them, and augment the schemas of views by keeping `derive-count' attributes on them as extra information.