Upload
amberly-nash
View
213
Download
0
Tags:
Embed Size (px)
Citation preview
1
1
Improving interoperability in StatisticsImproving interoperability in StatisticsSome considerations on the impact of SDMXSome considerations on the impact of SDMX
59th Plenary of the CES
Geneva, 14 June 2011
Rune GløersenIT DirectorStatistics Norway
2
Contents
• The characteristics of processes and data at NSIs
• Applicable standards for various business processes
• The preconditions for increased interoperability
• A top-down approach to further standardisation
• SDMX as part of the industrialisation of statistics
5
Specifyneeds
Design Build Evaluate
Quality Management/Metadata Management
Process stages and data archiving
Data archiving spans the 4 main business processes,and comprises 4 steady states of the data life cycle
6
Dissemination of aggregated statistics using SDMX
SDMXConversion
SDMX Common Architecture
Can (somewhat) easily be streamlined
7
Dissemination of any statistical data using SDMX
SDMXConv
SDMX Common Architecture
Requires a paramount strategy
8
Specifyneeds
Design Build Evaluate
Quality Management/Metadata Management
Adopting standards
DDISDMX
?
9
The diversity of users, needs and data flows
Public
Domain specific
Research
Questionnaires
Data transfersRegisters
Common high level models, vocabulary etc
10
Challenges
• The high-level decision to use SDMX for the exchange of statistical data; how should this be envisaged?
– The role of the standardisation experts, the IT experts, the subject domain experts and the top management
• SDMX implementation is strategic, but is regarded as technical– The importance and impact of the Information Model and the Metadata
Common Vocabulary
• Choosing standards; DDI, SDMX, DSPL etc.– No standard is likely to fit all purposes.– Will a common high-level information model contribute to easier
implementation of standards?– Can a high-level information model bridge different standards?
• Provide well defined interfaces, or develop software to hide the challenges?
– Common requirements for the quality of software
11
Improved interoperability Some trends
Organisationalinteroperability
Technologicalinteroperability
Semanticalinteroperability
12
Maturity growth in e-Government
OrganisationalInteroperability
SemanticalInteroperabilitySource: www.semicolon.no
Analytical Framework for e-Government Interoperability
SharingKnowledge
Aligning WorkProcesses
Joining ValueCreation
AligningStrategies
Bilateral data exchange, semi automated,Technical specifications and standards
Share best practises, metadata specifications,Set up standards for technical systems and dataexchange
Common information models, process models and service catalogues, shared development costs
Legislation,Whatever
13
Common GenericIndustrial Statistics
GSBPM GSIM
Methods Technology
Statistical Concepts Information Concepts
Statistical HowTo Production HowTo
conc
eptu
alpr
actic
alIndustrializing Statistics
De-coupling content and technical standardisation
14
Conclusions
• Standardisation is not a goal in itself; any standardisation effort must be based on well defined business cases. Success requires a top-down, management driven approach.
• The adoption of SDMX must be aligned with the on-going process oriented developments among NSI’s.
• Utilize the benefits of SDMX for the exchange of aggregated data, improve the international harmonisation of requirements, and simplify implementation whenever possible.
• Agreeing on common high-level models, creates an opportunity for flexible, targeted and effective solutions on the detailed level, still harmonised within a standardised framework
• The statistical community should act as an industry, not only as individuals, in order to increase commercial attention to the industry of statistics
15
Actions ?
• (Continue to) set up a common reference framework comprising the objectives of harmonisation/standardisation
– Appreciate clustered initiatives, but require precise description on the contributions to the overall objectives
– Better prioritisation among projects; it is unlikely that we can achieve all goals at once
– Improve governance and coordination– Let the drivers drive– Evaluate
• Decide where to provide for best practises, architectures, standards and tools/shared software components
• Improve the strategy on how to coordinate process developments with subject matter/domain specific developments
• Provide for innovation