Towards a normalised, domain-independent model for modelling the contents of statistical data and associated metadata Or: How to design correct and globally consistent Data Structure Definitions Bo Sundgren, Statistics Sweden OECD Expert Group Meeting on Statistical Data and Metadata eXchange

Contents By Example (based on a simple generic model) Actors Utilities Complex objects

UNESCO model version 1 (to be revised)

Everything clickable OBJECT VARIABLE Lefthand click Righthand click Select: - object - variable Retrieve metadata: - definition - value set, classification - questionnaire - quality declaration - survey documentation

Our propositions The statistical data/metadata model presented here is general and domain-independent. It will cover all kinds of data and metadata to be made publicly available on the Internet and to be exchanged between national statistical agencies and international organisations. This proposition has been verified in a number of cases. So far the proposition has not been falsified in any case. This generic model can be transformed in a systematic way into an SDMX-compliant generic model expressed in XML. Since cube models, as actually practiced in national statistical agencies and international organisations, differ slightly between themselves and cannot always be said to be standardised, we propose the transformation to take place in two steps: –Step1: Non-standardised cubes are transformed into normalised cubes as defined here. –Step 2: Normalised cubes are transformed into standardised SDMX cubes (to be defined).