Designing fit for purpose data structures
Download as PPTX, PDF2 likes148 views
The workshop led by Stephen Bounds addresses the complexities and tensions involved in designing effective data structures, focusing on information value, access versus capture trade-offs, and the layering of processes and practices. It explores concepts like long-term versus short-term information usefulness, the psychology of information perception, and methodologies for value assessment and data enrichment. Key discussions include the importance of recordkeeping and archiving in maintaining data integrity and compliance with privacy regulations.
Designing fit for purpose data structures
- 1. Designing fit for purpose data structures Stephen Bounds Director and PrincipalConsultant
- 2. About me • 20 years experience in IM, IT & KM • 5 years running KnowQuestion • 10 years as a Federal public servant • Last public service job Deputy CIO, Information Management at Australian Communications and Media Authority (ACMA)
- 3. Today’s workshop • Tensions in data structure choices • Information value over time • Access / capture trade off • Process / practice layering • Recordkeeping and archiving
- 5. The four tensions Access Usefulness Preservation Privacy Long term considerations Short term considerations
- 6. How is information useful? For me For my team For my organisation n For the public public
- 7. When is it useful? Now Next Later Eventually
- 8. Psychological distance We think more concretely about things that are nearer to us, and more abstractly about things further away Temporal distance Whether an event is a long time in the future or the past Spatial distance Whether you are separated by a large geographic distance Social distance Whether you feel you know someone well or not Hypothetical distance Whether you think an event is likely to occur or not
- 9. Relative information value 0 2 4 6 8 10 Me Team Organisation Everyone
- 10. Perceived value over time 0 2 4 6 8 10 Now Soon Later Eventually
- 11. Reused information over time 0 20 40 60 80 100 Now Soon Later Eventually %reused
- 12. Value of reused information 0 2 4 6 8 10 Now Soon Later Eventually
- 13. Access / capture costs Costs of … Less structured More structured Capture Lower Higher Access Higher Lower
- 14. Impact of trade off Cost of capture Cost of access
- 15. Exercise 1:Value assessment A social worker has to fill out a case notes record every time they visit a client. They normally take handwritten notes and then transcribe into the system back in the office. • Assess each field on the basis of: • Scope of information value • Change in information value over time • Capture / access trade offs being made
- 16. Exercise 1:Value assessment • It is natural for people to discount the need for information that is of less direct value to them • How do you get buy-in when capturing information that is not directly relevant?
- 17. Key design problem • Case work is uncertain • Results of any case hugely variable • Subject to wide range of unpredictable factors • With a large volume of cases, it is possible to establish if systems are producing better results more of the time • Propensity inclination to behave in a particular way
- 18. Value assessment No long term tracking tracking Pr Case cost ($) Cost ($) Minimal long term impact 0.5 $80 $0 Significant long term impact 0.45 $80 $2,000 Critical long term impact 0.05 $80 $25,000
- 19. Value assessment Long term tracking Pr Case cost ($) Cost ($) Minimal long term impact 0.6 $130 $0 Significant long term impact 0.37 $130 $2,000 Critical long term impact 0.03 $130 $25,000
- 20. Monte Carlo simulations • Monte Carlo simulations build a model of a world and use this to simulate outcomes. • Re-running the model multiple times shows both variability and expected (average) results. • Can approximate real-world results without a known algorithm, eg calculating area of a circle:
- 22. Exercise 2:Value perception An updated case file format has been implemented which adds additional structured information. Specifically, the new form now references people by ID, and categorises referral type and incident type • Case officers are pushing back on the extra overhead. • What alternative options exist? • Simulate a relative cost/benefit of each pathway • How do we validate our assumptions?
- 23. Exercise 2:Value perception Note about modelling • Assumptions are critical • Never hide these – key to informed consent
- 24. Process-practice layering • All organisations do it, but often not consciously thought about • Practice – free-form execution that respects the expertise of individuals and teams • Process – structured set of repeatable steps • Process “touch points” used within practice to provide accountability and evidence
- 28. Information serves different purposes Evidence = That decisions and actions were taken Rationale = Why decisions and actions were taken Type of record: objective, factual, explicit subjective, opinion, tacit
- 29. Very broadly … More structured information Less structured information
- 30. Common objection More structured information Less structured information Valuable BI
- 31. Options to enrich data • Get someone else to add the structure • Contemporaneous • Post hoc • Automatic post processing (eg text to speech), keyword matching, semantic analysis • Inferences and trends based on client clustering • Last (worst) resort: get the expert to do it
- 32. Exercise 3: Process / practice • Which parts of the case management workflow are process and/or practice? Why? • Where is data enrichment most necessary, and what is the business value? • How do you build a business case around changing the responsibility for data capture and enrichment?
- 33. Recordkeeping and archiving Long term data storage Records Archives Backups
- 34. 3 different, overlapping concepts Collection type Purpose Retention Records Evidence about a business process, both of actions and supporting decisions Kept until no longer needed Archives Information with intrinsic value outside of the business process that generated it Kept indefinitely Backups Redundant information that supports timely disaster recovery Kept as long as risk management requires it
- 35. Remember… Not all records are in an archive Not all archives contain records
- 36. Keep or destroy? • Archives Act 1983 has no proactive requirement to destroy – NAA approves retention rules for records • Privacy Act 1988 requires reasonable steps to destroy personal information once it is no longer required • Discussion point – • Why destroy information? • Why keep information?
- 37. Information collection lifecycle Commissioning Creation Maintenance/ Migration Disposal/ Loss Transferral/ Duplication
- 38. Archive management To unlock value of archive to users, key requirements are: • Identification – Persistent, actionable identifiers • Discoverability – Indexing, cataloguing, and access • Preservation – Integrity, authenticity, and readability • Decision support – Evaluating relevance and currency
- 39. Exercise 4: Information collections • Consider the listed collections of information • Which are records, archives and/or backups? • Which hold information relevant to Privacy Act? • Trace 2 of these information collections from commissioning through to final disposal.
- 40. Thank you!