Slowly changing dimensions specification: a relational algebra approach

Abstract

In real-world applications, data warehouses usually incorporate some dimension tables in their schemas that have the ability to change over time, reflecting modifications occurred in the respective information sources, involving some of their attributes. These dimension tables are commonly recognized as Slowly Changing Dimensions. Dealing with them is one of the most regular activities of a data warehouse's designer. It is crucial to ensure that the dimension's attributes changes be correctly reflected without affect the data warehouse consistency. For years, researchers and technicians have assumed several Slowly Changing Dimensions strategies using them as the most usual ways to reflect changing and keep historical data. In this paper we will discuss Slowly Changing Dimensions in general terms, presenting their main characteristics and problems, and suggesting a generic formal specification based on Relational Algebra to cover all their known types, with particular emphasis to type 4.