troy delbridge, private healthcare australia, focus day, presentation at chief data & analytics...
TRANSCRIPT
Data Governance & Focus Day
Iden3fying the Specific Requirements for Meta Data Management to Support Your Data
Governance Ini3a3ve
Dr Troy Delbridge Independent ICT consultant (former Chief Data & Informa8on Officer at Private Healthcare Australia)
What is ‘metadata’? • Metadata is data that describes or gives informa7on about other data. • It is informa7on about the contents and the context of your data is of
various types, e.g.; – Structural -‐ file format, size, media type. Books/documents -‐ pages, chapters,
tables of contents, indexes, glossaries, etc., – Descrip-ve -‐ describes an object for purposes of discovery and iden7fica7on -‐
such as author, 7tle, subject, etc., – Administra-ve/Technical/Opera-onal -‐ track how data is used -‐ encoding
and processed over 7me, a record of who and when they accessed data, a log of issues that exist with that data,
– Business -‐ what data actually means, and if it can be used in a par7cular business context -‐ not sta7c, as extending its use creates new meaning and rela7onships.(e.g. the rela7onship with reference data or the iden7ty of a master en7ty and its rela7onship to the current data.
• Some metadata is not about actual data, but is about analy7cal models and their components, visualiza7ons, and any assump7ons used.
Defini7ons of Metadata
Adapted from Baca ed. (2008) – Intro’ to Metadata
Examples • Documents, spreadsheets, mul7media, (videos, images, web-‐pages),
emails, voice calls/messages, databases, etc. – e.g. the ‘Proper7es’ tables of MS Office docs, Data(base) Dic7onaries, Glossaries,
etc. • In the pre-‐computer era -‐ book classifica7on and cataloguing systems;
– Dewey Decimal Classifica7on (1876), Lib. Of Congress (1897), etc. • Typically a glossary or data dic7onary for a database.
– These can oZen be just Excel spreadsheets or Word documents. • Electronic document &records management systems (EDRMS), XML
Schemas, etc.
• Mul7ple different metadata standards exist. – An good example in healthcare; METeOR (the metadata registry for health & welfare).
Metadata entry table -‐ Word document
METeOR metadata informa7on types
Source: AIHW 2016
Why you need metadata • Organises your data into a structured catalogue. • Loca7on informa7on -‐ where it is loaded on IT system/s, as well as it is
physically (hard-‐stores, servers and media type/s). • Ability to access metadata records allows records to be filtered and
searched using relevant criteria to; – iden7fy resources, – bring similar resources together, – dis7nguish dissimilar resources.
• Supports archiving & preserva7on of resources, • Provides for consistency between users & aids analysis. • Provenance and ownership. • Is essen7al to obtain a comprehensive understanding of an organiza7on's
data/informa7on assets. • Cri7cal to fully leverage your data/informa7on assets effec7vely in
Analy7cs/BI, Repor7ng and Opera7ons (both business and IT).
Ques7ons that should be asked about every dataset in use
Good data governance requires the following ques7ons be addressed: • Where does the data come from? • Is it the right data for its intended use? • Are you allowed to use the data, and if so, in what ways? • What other data choices could be made? • Was the data transformed, and if so, how? • Are there any quality problems? • Does everyone in your organisa7on understand the data in the same way?
If you have good metadata data you can answer most of these!
Metadata in a Governance context Metadata should be a key part of your data governance program, as it provides a master record of the all the data resources in your organisa7on;
• Know what data your organisa7on is collec7ng, • Provenance – from where the data is being sourced, and by whom?
– Determine who is collec7ng it, who has actual ownership/responsibility for the data (there may more than one owner), and who are the end users of the data.
• Assess and document the quality of your data (down to field level), • Understanding the data life cycle and where it is used in your business processes, • Authorisa7ons/security – who can use it, in what contexts, and how it is secured, • Iden7fying links between business units – from collec7on, to use, to storage/archiving, and
disposal, • A means of recording reference data changes – e.g. data field changes, classifica7on changes,
etc. which can have a significant effect on business opera7ons and require business process and IT system modifica7ons.
Where to start -‐ if star7ng from scratch • Will depend on the resources available and the size and nature of your
organisa7on's data assets; – Projects to implement whole of organisa7on metadata have been known to fail when
agempted as a large scale exercise requiring rapid comple7on, – A beger approach is to tackle it at the same 7me as IT and business improvement projects
are being undertaken – i.e. do it as part of the discovery and assessment phase in these projects,
– Requires that the data owners and the users be included, – Target priority business problems that require immediate agen7on from an IT/systems
perspec7ve – it may only be one or two small projects ini7ally. • Iden7fy all the data/informa7on sources relevant to a given problem – incl.
proposed IT system builds/upgrades, and business process improvements. – Record and asses the associated metadata for BAU, then analyse whether current the
process’s data, or other data, can support the goals of the project beger. • Expand your organisa7ons metadata coverage over 7me to obtain the eventual
end goal of having a holis7c understanding of all its data assets and how they are being used.
In Summary Good metadata and its management will be cri7cal to the long term success for your data governance ini7a7ves. • Oversight of the metadata associated with all of your data assets will
make the tasks of data collec7on, maintenance, integra7on/linkage, access/sharing, analysis, and QA that much easier.
• You will have the ability to see how your data is being collected and used in your business processes.
• You will understand your business processes and opera7onal workflows beger, and it will aid you in improving business processes and developing more efficient workflows.
• Will know the quality (validity, accuracy) of your data and the poten7al for its reuse in other processes and in other business units.
• You will know who your custodians are – who collects it, who is using it and for what purposes.
• Allow beger integra7on into your Analy7cs/BI and Repor7ng.
Dr Troy Delbridge E: [email protected] M: +61 0409 220 280
Some useful references on metadata Understanding Metadata, NISO (2004) (hgp://www.niso.org/publica7ons/press/UnderstandingMetadata.pdf) Introduc8on to Metadata, 3rd Edi8on (2016), The Ge]y Research Ins8tute -‐ hgp://www.gegy.edu/publica7ons/intrometadata/ OECD Glossary of Sta8s8cal Terms: “Metadata” (2005) via their web portal -‐ hgps://stats.oecd.org/glossary/detail.asp?ID=5136 METeOR (Metadata Online Registry), AIHW -‐ hgp://meteor.aihw.gov.au/content/index.phtml/itemId/181162