Abstract—Data schema represents the arrangement of fact
table and dimension tables and the relations between them. In
data warehouse development, selecting a right and appropriate
data schema (Snowflake, Star, Star Cluster …) has an
important Impact on performance and usability of the designed
data warehouse. One of the problems that exists in data
warehouse development is lack of a comprehensive and sound
selection framework to choose an appropriate schema for the
data warehouse at hand by considering application
domain-specific conditions. In this paper, we present a schema
selection framework that is based on a decision tree for solving
the problem of choosing right schema for a data warehouse. The
main selection criteria that are used in the presented decision
tree are query type, attribute type, dimension table type and
existence of index. To evaluate correctness and soundness of this
framework, we have designed a test bed that includes multiple
data warehouses and we have created all the possible states in
decision tree of schema selection framework. Then we designed
all types of queries and performed the designed queries on these
data warehouses. The results confirm the correct functionality
of the schema selection framework.
Index Terms—Data warehouse, framework, online
transaction processing, schema selection.
M. H. Peyravi is with Department Of Computer & Science, Sarvestan
Branch, Islamic Azad University, Fars, Iran (email:Peyravi@iausarv.ac.ir).
Cite:M. H. Peyravi, "A Schema Selection Framework for Data Warehouse Design," International Journal of Machine Learning and Computing vol.2, no. 3, pp. 222-225, 2012.