Representing and Utilizing DDI in Relational Databases

Abstract

This document is primarily intended for implementers of DDI-based metadata stores who are considering different technical options for housing and managing their metadata. The Data Documentation Initiative (DDI) metadata specification is expressed in the form of XML schema. With version 3, the DDI specification has become quite complex, including 21 namespaces and 846 elements7. Organizations employing DDI, or considering doing so, may want to 1. store and manage the metadata elements in relational databases, for reasons of integration with existing systems, familiarity with the concepts of relational databases (such as Structured Query Language), systems performance, and/or other reasons; 2. select only the subset of the available DDI metadata elements that is of utility to their work, and have the flexibility of capturing metadata they need that would not fit into the DDI model. This paper discusses advantages and disadvantages of the relational database approach to managing DDI. It also describes methods for modeling DDI in relational databases and for formally defining subsets of DDI to employ in this environment