group the base data along various dimensions, corresponding to different sets of group-by attributes, and compute various aggregate functions, often called measures.
As an example, the cube operator =-=[GBLP96]-=- can be used to define several such summary tables with one statement.
In Section 4.2 we show how the refresh function handles MIN and MAX aggregate functions.
2.3 Previous Aggregation Techniques The technique of =-=[GMS93]-=- works by computing a set of insertions and deletions (combined into one delta set with positive and negative counts) for each materialized view.
The concept of deriving minimal auxiliary views presented in this paper complements incremental maintenance t...
Hull and Zhou =-=[HZ96]-=- make views selfmaintainable by pushing down projections and selections to the base tables and storing these at the data warehouse. [QGMW96] present an algorithm for making views selfmain...
Citation Context ...uction to this section, a full examination of the issues involved in deriving maintenance expressions for GPSJ views in general is beyond the scope of this paper.
Generalized projection is an extension of duplicate-eliminating projection, where the schema A can include both aggregates and regular attributes.
Particularly important are materialized views that involve aggregation because dat... We use the generalized projection operator, \Pi A , to represent aggregation =-=[GHQ95]-=-.
All such tuples are then recomputed from the base tables.