Integrating Star and Snowflake Schemas in Data Warehouses

Abstract

A fundamental issue encountered by the research community of data warehouses (DWs) is the modeling of data. In this paper, a new design is proposed, named the starnest schema, for the logical modeling of DWs. Using nested methodology, data semantics can be explicitly represented. Part of the design involves providing a translation mechanism from the star/snowflake schemas to a nested representation. The novel schema proposed in this paper is accomplished by converting the fact-dimension schema to a fact-nested dimension schema. The transformation of the denormalized dimension tables to nested dimension tables increases the efficiency of query execution by reducing the number of tuples accessed for query retrieval since dimensional attributes can be used directly in the Group-by clause. In order to facilitate the implementation of the proposed approach, specific algorithms are built based on the starnest schema.