Approaching ETL processes specification using a pattern-based ontology

Abstract

The development of software projects is often based on the composition of components for creating new products and components through the promotion of reusable techniques. These pre-configured components are sometimes based on well-known and validated design-patterns describing abstract solutions for solving recurring problems. The data warehouse ETL development life cycle shares the main steps of most typical phases of any software process development. Considering that patterns have been broadly used in many software areas as a way to increase reliability, reduce development risks and enhance standards compliance, a pattern-oriented approach for the development of ETL systems can be achieve, providing a more flexible approach for ETL implementation. Appealing to an ontology specification, in this paper we present and discuss contextual data for describing ETL patterns based on their structural properties. The use of an ontology allows for the interpretation of ETL patterns by a computer and used posteriorly to rule its instantiation to physical models that can be executed using existing commercial tools.