[Steve] Related to this point, there have been some discussions recently about the GSBPM providing a framework for SDMX and DDI standards to work together within the statistical production process. The emerging consensus is that DDI may be the better standard for microdata, and SDMX for aggregate data. In practical terms for national statistical organisations, this means that the Collect and Process phases of the GSBPM would use DDI metadata formats, whilst Analyse and Disseminate would use SDMX. Design, Build and Archive would probably use both (depending on whether they are dealing with microdata or aggregates). For international organisations, SDMX is more likely to be used throughout. This approach is part of a proposed joint statement from DDI and SDMX administrators, which will hopefully be agreed in the next few months. The diagram below summarises this approach, but still needs to be refined (background information can be found in this paper) . Whilst this does not, of course, define metadata flows, it defines formats, which should help!
[Thérèse/Al] An example of some brainstorming on this topic from the ABS in August 2010
[Jenny Linnerud] Text mining GSBPM to identify IOCs (Information Object Candidates).
[Isabel Morgado] New version for comments: Draft example to present in METIS Workshop - Metadata Flows within GSBPM (Excel file) [Isabel Morgado] Paper for discussion: Metadata Flows within GSBPM (Excel file)