Data Warehouse, Scripting Language, Data Pump, Data Mining

The bottleneck of a data warehouse implementation is the ETL (extraction, transformation, and load) process, which carries out the initial population of the data warehouse and its further (usually periodical) updates. There is a number of software products supporting the OLAP analysis. However, the ETL process implementation is not repeatable in a significant way. This paper reports on a research of a model-based data transformation applicable to data warehouse population and updates. The ETL process is based on a metadata repository, which contains data models of the data sources, the target data warehouse model, and the correspondence among them