2.0 DDI 3 Design DDI 3 adds a lot of complexity, because it is designed to support the entire statistical lifecycle, rather than just the archival part. This places a major emphasis on being able to identify, version, and maintain the metadata throughout that process.
Further, it allows for groups of studies to be documented in relation to each other, for comparison purposes or to track versions as the metadata grows throughout the lifecycle.
Modularity supports both requirements by allowing a tighter focus on metadata that is of interest to a specific application or user. While this may seem complex, once the basic design is understood, it allows for a much more exact expression of the metadata, and, in the long term, better management and processing of that metadata.
2.1 Design Rules The demands of the changes noted above made it clear that DDI 3 needed to outline clear design rules to ensure consistency in the creation and development of DDI 3
• Persistent sections should be separate from dynamic information. What parts change when a data file moves from one “home” to another, or changes something like its physical storage structure? • Information modules should follow the various life cycle paths • Information used for discovery should be in non-specialized modules • Links should be unidirectional to avoid loops and broken links as materials are repacked or versioned • Links should point back in time with materials later in the lifecycle pointing to existing materials rather than going back and adding new links • All comparisons are pair wise, comparing a source with a target. • Groups inherit down the tree unless there is a clear local override provided • Functionality of DDI 2.1 would be preserved • Different types of XML elements will inherit from each other in XML schemes, to simplify programmatic processing of basic types which have many different variations throughout the lifecycle. • Metadata will be expressed in ways which support both human-readability and machine-processing.