Conceptual Modeling for Data Warehouse and OLAP Applications

Abstract

The advantages of using conceptual models for database design are well known. In particular, they facilitate the communication between users and designers since they do not require the knowledge of specific features of the underlying implementation platform. Further, schemas developed using conceptual models can be mapped to different logical models, such as the relational, objectrelational, or object-oriented models, thus simplifying technological changes. Finally, the logical model is translated into a physical one according to the underlying implementation platform. Nevertheless, the domain of conceptual modeling for data warehouse applications is still at a research stage. The current state of affairs is that logical models are used for designing data warehouses, i.e., using star and snowflake schemas in the relational model. These schemas provide a multidimensional view of data where measures (e.g., quantity of products sold) are analyzed from different perspectives or dimensions (e.g., by product) and at different levels of detail with the help of hierarchies. On-line analytical processing (OLAP) systems allow users to perform automatic aggregations of measures while traversing hierarchies: the roll-up operation transforms detailed measures into aggregated values (e.g., daily into monthly sales) while the drill-down operation does the contrary. Star and snowflake schemas have several disadvantages, such as the inclusion of implementation details and the inadequacy of representing different kinds of hierarchies existing in real-world applications. In order to facilitate users to express their analysis needs, it is necessary to represent data requirements for data warehouses at the conceptual level. A conceptual multidimensional model should provide a graphical support (Rizzi, 2007) and allow representing facts, measures, dimensions, and different kinds of hierarchies.