DDI 4 Core - Development of a Robust and Sustainable Model

Motivation

DDI has always faced the requirement of dealing with a large range of data, both for archival purposes and to provide support for the entire production lifecycle to large studies and statistical agencies. The result of this work is a model which in many important respects is domain-independent. Recent developments in the research world are placing a greater emphasis on cross-domain integration of data, and data coming from new sources, some of it in unfamiliar forms (e.g., “big data,” social media, sensor data, etc.). Social Science is no different from other domains – the DDI community is faced with a requirement for a more flexible ability to describe and manage, now available in a wider variety of forms.

The DDI 4 Core is intended to provide useful functionality in response to this requirement. In presentations, review comments, and discussions it has become apparent that some aspects of the DDI 4 model included in the DDI 4 Prototype are of especially high value. Identified features include the conceptual aspects of variables and classifications, the datum-oriented description of data, and the use of the process model to describe data lineage (the processing involved in the provenance of data), with a further emphasis on alignment with external standards and the use of a UML model as a primary deliverable.

The DDI 4 Core will include not only the XML and RDF syntax representations, but will also deliver the UML from which they are derived in the form of Canonical XMI, a portable, tool-independent expression of the model. This not only makes DDI available for representation in other syntaxes and systems, but provides a stable basis for the maintenance of the model into the future.