Digital Curation 101: Preserve

Digital Curation 101: Preserve Michael Day UKOLN, University of Bath [email_address]

Presentation outline:
Preservation in the curation life-cycle
Roles and responsibilities
Reasons for preserving research data
Digital preservation challenges and strategies
Major types of research data collection
Infrastructures for preservation and curation

Module outline
This module will explore actions required to ensure long-term preservation and retention of the authoritative nature of data
Preservation actions should ensure that data remains authentic, reliable and usable while maintaining its integrity.
Actions include data cleaning, validation, assigning preservation metadata, assigning representation information and ensuring acceptable data structures or file formats

Learning outcomes
An greater awareness of the factors that need to be taken into account when considering how to preserve research data (and other materials) over time
A deeper understanding of the preservation options currently available

Preservation in the DCC lifecycle
In the DCC Curation Lifecycle Model, the “Preservation Action” stage:
Immediately follows the “Ingest” stage
Is followed by the “Store” stage
Is directly linked with “Transform” and “Appraise and Select” stages
Includes major elements from the inner circles:
Description and Representation Information, Preservation Planning, Community Watch

Preservation in the DCC lifecycle
There are major dependencies on the rest of the curation process
The creation stage is normally the best time to ensure that data are fit-for-purpose and “preservable”
Need to document both explicit and implicit knowledge, contexts (part of the metadata issue)
Preservation Planning informs ingest strategies as well as preservation actions and transformations

Who undertakes preservation?
Researchers
Indirectly - they have most direct contact with creation stage, and understand how data can be used
Directly - sometimes responsible for maintaining community data collections
Information professionals
Sometimes, but it depends on the context
IT professionals
Primarily informaticians working with scientists

Roles and responsibilities (1)
Long-lived data collections (NSB)
Data authors
Data managers
Data scientists
Data users
Funding agencies
Dealing with data (JISC)
Scientist
Institution
Data centre
User
Funder
Publisher

Scientists
Initial creation and use of data
Expectation of first use and in gaining appropriate credit and recognition
Responsible for:
Managing data for life of project
For using standards (where possible)
For complying with data policies
For making the data available in a form that can (easily?) be used by others

Institutions:
Role less clear
Institutional policies may require short-term management of data
Advocacy and training
Some institutions are developing repository services
Are rarely currently used for research data
Federated approaches maintain disciplinary involvement

Data centres
Undertakes curation and provides access
Responsible for:
Selection and ingest
Participating in the development of standards
Protecting the rights of data creators
Supporting ingest and metadata capture
Supporting re-use (tools and services)
Training

Users:
Users of third-party data
Responsible for:
Adhering to any licenses and restrictions on use
Acknowledging data creators and curators
Managing any derived data
Provide feedback to scientists and data centres

Funding bodies:
Acting at policy level
Responsible for:
Considering wider policy perspectives
Developing policies in co-operation with other stakeholders
Monitoring and enforcing data policies
Support for long-term data management
Support for data curation

What is research data?
An extremely broad category of material:
“... any information that can be stored in digital form, including text, numbers, images, video or movies, audio, software, algorithms, equations, animations, models, simulations, etc.” (National Science Board, Long-lived digital data collections, 2005)
In practice, it can mean almost anything

Why curate research data? (1)
Part of the normal research process:
The need for others to validate and replicate research
In some disciplines, supporting data is routinely made available to reviewers and linked from journal papers
Principles of sharing and openness are firmly embedded in some disciplines

Extrinsic and intrinsic value;
High investment in research
Data can be very expensive to capture and analyse
Data is impossible to recreate once lost
Observational data (by definition) is irreplaceable
Current generations of instruments can gather more data than can be analysed

The potential for creating 'new' knowledge from existing data:
Re-use, re-analysis, data mining
Annotation, e.g. in molecular biology astronomy
Combining datasets in innovative ways, e.g. mapping biodiversity data onto ecological GIS
“Science 2.0”

It is increasingly a requirement of some research funding bodies
Some have quite mature data retention policies (not necessarily for permanent retention)
Increasing expectation of access to data from publicly-funded research
OECD Principles and guidelines for access to research data from public funding (2007)

Institutional asset management:
Universities and other research organisations invest very large sums of money into research activities
Research data is a key output of this activity
It is, therefore, an institutional asset that needs stewardship

Promoting the institution, research group or individual:
Re-use helps promote visibility and 'impact'
Institutions become acknowledged 'centres of competence'

Preservation challenges (1)
Media (1)
Currently magnetic or optical tape and disks, some devices (e.g., memory sticks)
Examples include: CD, DVD (optical), DAT, DLT, laptop hard drives (magnetic)
Unknown lifetimes
Subject to differences in quality or storage conditions
But relatively short lifetimes compared to paper or good quality microform
Lifetimes measured in years rather than decades

Media (2)
Technical solutions
Longer lasting media:
e.g. Norsam's High Density Rosetta system - analogue storage on nickel plates
COM (output to good-quality microform)
Keeping paper copies!
Periodic copying of data bits on to new media (refreshing) - data management solution
Principle of active management

Hardware and software dependence
Most digital objects are dependent on particular configurations of hardware and software
Relatively short obsolescence cycles for:
Hardware
Scientific instrumentation, peripherals (e.g. floppy disk drives)
Software
e.g., word-processing files, CAD

Conceptual problems (1)
What is an digital object?
Some are analogues of traditional objects, e.g. meeting minutes, research papers
Others are not, e.g. Web pages, GIS, 3D models of chemical structures
Complexity
Dynamic nature

Three layers:
Physical: the bits stored on a particular medium
Logical: defines how the bits are used by a software application, based on data types (e.g. ASCII); in order to understand (or preserve) the bits, we need to know how to process this
Conceptual: things that we deal with in the real world
From: Ken Thibodeau, “Overview of technological approaches to digital preservation and challenges in coming years.” In: The state of digital preservation: an international perspective. CLIR, 2002. http://www.clir.org/

On which of these layers should preservation activities focus?
We need to preserve the ability to reproduce the objects, not just the bits
In fact, we can change the bits and logical representation and still reproduce an authentic conceptual object (e.g. converting into PDF)
Authenticity and integrity
How can we trust that an object is what it claims to be?
Digital information can easily be changed by accident or design

Some general principles (1)
Most of the technical problems associated with long-term digital preservation can be solved if a life-cycle management approach is adopted
i.e. a continual programme of active management
Ideally, combines both managerial and technical processes, e.g., as in the OAIS Model
Many current systems are attempting to support this approach
Preservation strategies need to be seen in this wider context
Preservation needs to be considered at a very early stage in an object's life-cycle

There is a need to identify 'significant properties'
Recognises that preservation is context dependent
Helps with choosing an acceptable preservation strategy
Consider encapsulation
Surrounding the digital object - at least conceptually - with all of the information needed to decode and understand it (including software)
Produces autonomous 'self-describing' objects, reduces external dependencies (linked to the Information Package concept in the OAIS Reference Model)
Keep the original byte-stream

Metadata and documentation is vitally important
Relates to the OAIS concepts like Representation Information and Preservation Description Information
Functions
Records scientific meaning
Records the research context
Enables the development of finding aids
Standards are being developed that support digital preservation activities (e.g., the PREMIS Data Dictionary)

Digital preservation strategies
Three main families:
Technology preservation
Technology emulation
Information migration
Also:
Digital archaeology (rescue)

Technology preservation
The preservation of an information object together with all of the hardware and software needed to interpret it
Successfully preserves the look, feel and behaviour of the whole system (at least while the hardware and software still functions)
May have a role for historically important hardware
Severe problems with storage and ongoing maintenance, missing documentation
Would inevitably lead to 'museums' of “ageing and incompatible computer hardware” -- Mary Feeney
May have a shorter-term role for supporting the rescue of digital objects (digital archaeology)

Technology emulation (1)
Preserving the original bit-streams and application software; running this on emulator programs that mimic the behaviour of obsolete hardware
Emulators change over time
Chaining, rehosting
Emulation Virtual Machines
Running emulators on simplified 'virtual machines' that can be run on a range of different platforms
Virtual machines are migrated so the original bit-streams do not have to be

Benefits:
Technique already widely used, e.g. for emulating different hardware, computer games
Preserves (and uses) the original bits
Reduces the need for regular object transformations (but emulators and virtual machines may themselves need to be migrated)
Retains ‘look-and-feel’
May be the only approach possible where objects are complex or dependent on executable code
Less 'understanding' of formats is needed; little incremental cost in keeping additional formats

Challenges:
Do organisations have the technical skills necessary to implement the strategy?
Preserving 'look and feel' may not be needed for all objects
It will be difficult to know definitively whether user experience has been accurately preserved
Conclusions:
Promising family of approaches
Needs further practical application and research, e.g.
Dioscuri software (National Library of the Netherlands (KB), Nationaal Archief and Planets project)

Information migration (1)
Managed transformations:
A set of organised tasks designed to achieve the periodic transfer of digital information from one hardware and software configuration to another, or from one generation of computer technology to a subsequent one - CPA/RLG report (1996)
Abandons attempts to keep old technology (or substitutes for it) working
A 'known' solution used by data archives and software vendors (e.g., a linear migration strategy is used by software vendors for some data types, e.g. Microsoft Office files)
Focuses on the content (or properties) of objects

Main types (from OAIS Model):
Refreshment
Replication
Repackaging
Transformation
Challenges:
Labour intensive
There can be problems with ensuring the 'integrity and authenticity' of objects
Transformations need to be documented (part of the preservation metadata)

Uses:
Seems to be most suitable for dealing with large collections of similar objects
Migration can often be combined with some form of standardisation process, e.g., on ingest
ASCII
Bit-mapped-page images
Well-defined XML formats
Some variations: migration on Request (CAMiLEON project)
Keep original bits, migrate the rendering tools

Digital archaeology
Not so much a preservation strategy, but the default situation if there isn't one
Using various techniques to recover digital content from obsolete or damaged physical objects (media, hardware, etc.)
A time consuming process, needs specialised equipment and (in most cases) adequate documentation
Considered to be expensive (and risky)
Remains an option for content deemed to be of value

Choosing a strategy
Preservation strategies are not in competition (different strategies will work together)
It has been suggested that we should keep the original bits (with some documentation) in any case
But the strategy chosen has implications for:
The technical infrastructure required (and metadata)
Collection management priorities
Rights management
e.g, Owning the rights to re-engineer software
Costs
Planets project - PLATO preservation planning tool
Decision support tool

File formats and preservation
Formats can be identified and validated at ingest
JHOVE, PRONOM-DROID
Standardisation on ingest
Perceived wisdom suggests the adoption of open or non-proprietary standards, e.g. databases structured in XML, uncompressed images
However, we need more empirical data on how robust some of these standards are to random bit-rot

Other preservation challenges
Scale (1):
The “digital deluge”
e-Science
New generations of instruments
Computer simulations
Many terabytes generated per day, petabyte scale computing (and growing)
Cory Doctorow, “Welcome to the petacentre.” Nature , 455, pp 17-21, 4 Sep 2008

Scale (2):
Problems of scale are particularly acute in traditional 'big-science' disciplines:
Particle physics (e.g., the Large Hadron Collider)
Astronomy (sky surveys, etc)
But “smaller experiments will grow the fastest” (Szalay & Gray, Nature , 440, 413-4, 23 Mar 2006)
Bioinformatics, crystallography, engineering design, and many others
In some cases it may be cheaper just to generate the data again, e.g. for computer simulations

Complexity (1)
Research data is extremely diverse - not really a single category of material
tabular data, images, GIS, etc.
raw machine output vs, derived data
varying levels of structure (XML, legacy formats, etc.)
many different standards
Research data is not homogeneous
No one-size-fits-all approach possible

Complexity (2):
Even wider range of social contexts in which data is used (and shared)
DCC SCARP project has been exploring disciplinary factors in curation practice
Practice even within single disciplines is very fragmented
Case studies ongoing
Big-science archives, medical and social sciences, architecutre and engineering, biological images

Diverse research cultures
Data practices vary widely, even within a single discipline
Gene sequence data is typically deposited in public databases
In proteomics sharing is not so widespread; partly driven by lack of standards, but there is also concern about who have exploitation rights
Role of commercial interests
Pharmaceuticals, architecture and engineering, geological prospecting

Costs
Recent JISC study (2008) - focusing on the institution level
Some findings:
The complex service requirements for curating research data means that institutions are setting-up federated approaches to repository development
Currently ingest costs are much higher than long-term storage and preservation costs
Start-up (and R&D) costs are high, but there can be economies of scale

Research data collections (1)
A typology (1):
From National Science Board report Long-lived digital data collections (2005)
Research data collections – the products of one or more focused research projects
Resource or community data collections – collections that emerge to serve particular subject sub-disciplines
Reference data collections – serve a broader and more diverse set of user communities

Research data collections (2)
Data in “research data collections” is most at risk
A modern version of the “file-drawer problem”
Data stored on personal hard-drives or on media; largely undocumented
Particular challenge when the data creator has retired or moved to another institution
Data creators not aways aware of its potential value
The reward structure of science is not always helpful

Curation infrastructures (1)
Focus on the generic:
Need for a balance between:
The 'bottom-up' discipline-based drivers that promote the generation of research data
The policy level, looking to make cost effective investment in curation
When building Infrastructures, focus on the generic
Storage systems and middleware
Preservation services
Identifying the needs of the wider community

Curation infrastructures (2)
The need for collaboration:
Need for 'deep-infrastructure' recognised as far back as 1996 by the Task Force on Archiving of Digital Information
Digital preservation involves the "grander problem of organizing ourselves over time and as a society ... [to manoeuvre] effectively in a digital landscape" (p. 7)

Summing-up
Long-term preservation of research data is a big ongoing challenge
Solutions are based on the active management of data
Decisions needed on whether to adopt standard formats, identifying significant properties, preservation planning
Research disciplines and sub-disciplines are at different stages of maturity

The Future ...
“It is always a mistake for a historian to try and predict the future. Life, unlike science, is simply too full of surprises” - Richard J. Evans, In defence of history (1997, p. 62)

Further reading
National Science Board, Long-lived digital data collections: enabling research and education in the 21st century (NSF, 2005) http//www.nsf.gov/pubs/2005/nsb0540/
Liz Lyon, Dealing with data; roles, rights, responsibilities and relationships (JISC, 2007) http://www.jisc.ac.uk/whatwedo/programmes/digitalrepositories2005/dealingwithdata.aspx
Neil Beagrie, Jullia Chruszcz, and Brian Lavoie, Keeping research data safe: a cost model and guidance for UK universities (JISC, 2008) http://www.jisc.ac.uk/publications/publications/keepingresearchdatasafe.aspx

Thank you for your attention! “ Pigabyte” King Bladud’s Pigs in Bath (public art project), Summer 2008 http://www.kingbladudspigs.org/

Acknowledgements
UKOLN is funded by the Museums, Libraries and Archives Council (MLA), the Joint Information Systems Committee (JISC) of the UK higher and further education funding councils, as well as by project funding from the JISC, the European Union, and other sources. UKOLN also receives support from the University of Bath, where it is based.
More information: http://www.ukoln.ac.uk/

Digital Curation 101: Preserve

by Michael Day on Oct 08, 2008

Accessibility

Categories

Tags

Upload Details

Usage Rights

1 Embed 3

Statistics

Digital Curation 101: Preserve — Presentation Transcript