A Proposed Technique For Metadata Management For Data Warehouse

Abstract

Consistent metadata should be captured and stored in a repository to be shared by various users groups and software components. Users need the repository as a consistent documentation in order to effectively and efficiently achieve their particular tasks. On the other hand, software components may produce and consume the information of the repository. Thus, the repository has to provide a structure for fitting its utilization purposes. At the conceptual level, the model of metadata describes the structure of the repository. Each repository should provide a documented and user-extendable model, which easily allows the extension of the system for additional requirements. Metadata has been identified as a key factor on data warehouse projects. It captures all kind of information necessary to analyze, design, build, use and interpret the data warehouse contents. Standardization of metadata representation and exchange is of great importance in order to spread the use of metadata to enable the interoperability between repositories, and tool integration within data warehousing architectures. In this work

• We compared two standard models for metadata management, Open Information Model (OIM) and Common Warehouse Model (CWM). In this thesis we preferred to apply CWM due to: 1- OIM is much broader in scope than CWM, OIM is not specific for data warehouse only, but CWM deals more with metadata for data warehousing and provides a framework representing and exchanging metadata. 2- CWM is supported with the metadata object facility (MOF) standard for metadata exchange, whereas OIM is not (i.e. OIM is only descriptive model). 3- OIM supports relational-to-relational mappings only, while CWM is not tied to a particular model.