1. Topic 2 (ii) Metadata concepts, standards, models and registries
Questions from opening session
Common terminology – SDMX’s MCV, definitions for terms such as survey
Classification / typology of statistical metadata

2. Topic 2 (ii) Metadata concepts, standards, models and registries
Metadata classifications
Multidimensional classification for statistical metadata – by purpose (who, why), by content (what), by source and by form (how)
Metadata by usage or purpose
All processes produce metadata
metadata X each process – paradata (peridata)
Metadata by content – object attachment
Corporate metadata repository
Process, data, conceptual object, collection source
Finally combines –source X use

3. Metadata classifications (2)
Do we need a metadata classification by form – coded, structured, organization ?
SDMX requirements
Structural metadata – descriptors of the data (Bo’s Table 2c)
Reference metadata – contents and quality of data (Bo’s tables 2a, 2b and 2d)
Archive metadata
Administrative – Bo’s table 2b
Structure – Bo’s table 2c
Survey and definitional – Bo’s tables 2a, 2b, 2d

4. Metadata registries and harmonized content
Metadata classifications and standards – How do we move to more harmonization of metadata standards and classifications?
ESS – metadata are documented according to a standard metadata
SDMX metadata standards – good start – should definitions be adopted as part of METIS CMF?
Role of interoperability – should they be part of the CMF?
PC AXIS – SDMX
SDMX – 11179
other mappings / crosswalks

5. Data semantics and interoperability (data and semantics)
Two papers – US paper on data and datatypes; Statistics Netherlands on metadata models (classifications and datasets)
Data semantics – better understanding of the meaning of data (paragraph 2 US and paragraph 9 NLD)
Synonymous information and equality
Mechanisms needed to identify metadata items that are the same – synonyms
For categorical data – equality is defined the same way as for numbers - semantics of values must be the same to allow for comparisons
Are these ideas similar?

6. Statistical classifications
Logic of classification systems is essentially Boolean algebra – new view of classifications
Related view ontological systems
Figures 5 and 6 – Is the rule the classes must be mutual exclusive not respected?
Conclusion: formal semantics can play a constructive role in the development of models for statistical metadata
How does this view relate to Dan’s – ontology is formal means for organizing data and data descriptions