Abstract

A Data Warehouse (DW) is the infrastructural data foundation for Business Intelligence (BI) systems. This study proposes a methodology for designing a database (DB) schema for a DW. The core of the proposed methodology is a source-to-target schema conversion solution, based on directed-graph representation of a relational DB schema. It converts the graph representation of a source DB (e.g., one that supports operational information system) to a schema that would better fit DW requirements (a.k.a, "Star Schema"). The methodology does not aim at fully-automated conversion, but rather permits expert-user intervention for handling schema-design decisions that would require in-depth understanding of business context and interpretation. This manuscript presents the methodology foundations - graph-representation of a relational DB schema, and the schema conversion process. It also describes a prototype implementation of the proposed methodology, and discusses direction for future research progress.

Share

COinS