Object Role Modeling Enabled Metadata Repository

Abstract

The classic definition of metadata is simply stated as data about data. Straightforward as it may be, this data can be very powerful. In this paper we will explore the possibility of how we can leverage metadata in a typical enterprise environment to produce efficiencies, increase data quality/confidence and improve business decision making. To set the stage we will briefly describe the environment depicted in the diagram below. Transactional systems will act as the source of data. As data flows down stream it often resides in a data warehouse, which supports Decision Support Systems (DSS). A data warehouse provides an analytical environment and acts as a repository for an organization’s historical data. From the warehouse the data lineage is continued as data marts are created down stream for specific reporting areas and as integration is built among the different analytical environments. The collection of data lineages, we will refer to as the Information Supply Chain (ISC). Metadata is the key enabler binding together the web of data in the ISC. This type of environment often presents a challenge - business users and system analysts struggle trying to fully understand the business rules, data lineage and business definitions of data terminologies. We propose using Object Role Modeling (ORM) to design and build a central metadata repository that supports the retention of data definitions and data lineage of the ISC plus other metadata properties required in a repository system to play an active role throughout the lifecycle of data from source to BI reporting. The metadata repository will hold the knowledge of the enterprise shareable systems and data and make it available at the conceptual level. We will leverage ORM and the centralized repository to produce the following key features:

• Model and build a metadata management system

• Extract metadata from ORM into the metadata repository

• Explore metadata based automation of data mart schemas and ETL specifications

• Automate the code generation of stored procedures and class objects based on metadata