P1246990952BuZjz

1
Investigating Metadata for Long-Lived Geospatial
Resources An Exploration By Nancy J.
Hoebelheinrich Metadata Coordinator Digital
Library Systems Services
Stanford Digital Repository
2
To Be Discussed

• The Study
• Question asked
• Methodology used
• MD standards strengths / weaknesses
• Conclusions Recommendations
• Future work needed

3
The Study

• NGDA Project
• Pertinent project objective
• Collect and archive major segments of at-risk
  digital geospatial data and images
• Partners / Backgrounds / Areas of experience
• UCSB Alexandria Digital Library / Presentation
• Stanford Libraries Stanford Digital Repository
  / Preservation
• Differences in experiences gave rise to study
  question

4
Study Question

• What metadata is needed for long-lived geospatial
  data formats?
• Grounded in previous studies
• Hunolt paper for USGCRP Office
• Digital Preservation Coalition (UK)
• NSF
• OAIS Reference model
• Duerr, Parsons articles
• OCLC / RLG Preservation studies

5
Methodology used

• Evaluate four fairly typical geospatial data
  formats
• Shapefiles, DOQQs, DRGs Landsat 7 satellite
  images (preliminary)
• Compare / contrast 3 different approaches to
  documenting
• FGDC Content Standard
• CIESIN Geospatial Electronic Records
• PREMIS

6
Categories of information about the resources

• Environment (computer platforms)
• Semantic Underpinnings
• Domain specific terminology
• Provenance
• Data trustworthiness
• Data quality
• Appropriate use

7
Environment (computing platform)

• Definition characteristics of the hw / sw
  configuration that allow a resource to function
  properly
• Function could be defined as
• Rendering
• Viewing
• Using
• May need to be repeated

8
Environment, cont.

• All 3 systems have means for documenting these
  characteristics
• Both PREMIS and GER provide more granularity
  parsability, e.g., creatingApplication, sw, hw
  name, versions dependencies, environment type,
  etc.
• FGDC uses technical prerequisites native
  data set

9
Semantic Underpinnings

• Detailed concepts
• Meaning or essence of data
• Significance of data, i.e., why preserve it?
• Purpose or function served by data
• Intended community
• FGDC GER have fairly extensive set,
  particularly GER
• PREMIS NOT descriptive or domain specific, so
  not covered

10
Domain specific terminology

• For geospatial, particularly valuable
• Keywords associated with data themes
• Spatial coverage
• Time period
• Stratum coverage / place names
• GER FGDC cover
• PREMIS NOT descriptive MD

11
Provenance

• Detailed concepts
• Info about the events, parameters source data
  associated w/ construction of data set prior to
  ingestion
• Source of data
• Changes made to data inside the preservation
  archive
• FGDC, GER, PREMIS all ok for 1st 2
• FGDC NOT for last

12
Provenance, cont.

• Greater level granularity / parsability in
  PREMIS using Object, Event Agent entities
• See Example for Rumsey Historical Map Image
  Collection about descriptive MD transformation

13
Use of PREMIS Event Data Elements

• Example Event 1
• Transform of descriptive MD from MS Access db
  XML MODS

• Why this event?
• In case of questions from outside data provider
• Retain singular scripts transform mechanisms

14
PREMIS Event Excerpt (v1.1)
15
Use of PREMIS Event Data Elements

• Example Event 2
• Merge c\temp\states1c\temp \states2
  c\temp\USA
• (includes process merge and data sources
• Advantage can describe events once in
  repository, unlike FGDC, but
• Can include if prior to ingestion?

• Why this event?
• Important to describe processes during different
  phases of lifecycle, even prior to ingestion
• Not to be able to do so problemmatic for
  geospatial resources
• Is best practice issue for this domain

16
Data trustworthiness

• Detailed concepts
• Who are parties responsible for creation,
  development, storage, maintenance of data set
• Where is data located
• How is data available
• What important factors about the data should be
  preserved

17
Data trustworthiness, cont.

• Parties
• Location of data
• Factors to preserve

• FGDC, only originator, GER PREMIS more
  granular parsable
• GER FGDC seem more specific less inclusive
  (only POV of distributor) for last 2

18
Data Quality

• Detailed concepts
• General condition statement
• Accuracy of the data
• Fidelity of relationships within the data set
• Accuracy of measurements of the data
• FGDC has tags, but are very specific
• GER not much coverage here

19
Appropriate use

• Detailed concepts
• Legal use and liability statements
• Technical characteristics that impact use

• FGDC PREMIS have, GER NOT
• FGDC NOT, GER PREMIS have means of linking to
  format registry info
• More about format registries, later

20
PREMIS Significant properties

• Way within PREMIS to document
• Data trustworthiness data creator / provider
  reliable authentic
• Data quality describing completeness, logical
  consistency, attribute accuracy
• Data Provenance processes sources for dataset
  understandable reliable
• Appropriate use understanding of the specific
  needs of the designated community?
• Other important factors to preserve
• More work needs to be done in this area

21
Strengths weaknesses FGDC

• Rich in detail
• Specificity for the geospatial domain
• Ubiquity

• Very complex laborious to complete
• Poor means for describing relationships among
  file components of a digital resource
• No way to describe digital resource once within
  preservation archive

22
Strengths weaknesses GER

• Focus on archiving
• Comprehensive

• Little known as yet
• No data dictionary, so unclear how to apply tags
  (cardinality, repeatability, etc.)
• Relational DB format
• Unclear if and/or how to describe digital
  resource once within preservation archive

23
Strengths, weaknesses PREMIS

• Applicable at many levels of digital resource
  abstract physical
• Capability for describing relationships among
  file components of digital resource
• Capability for describing digital resource during
  its entire lifecycle within the preservation
  archive
• Generic focused upon preservation

• Not specific enough for geospatial
• Does not include critical semantics or
  descriptive information important for using
  digital resource
• Fairly young specification unclear how to
  document significant properties for digital
  resources

24
Recommendations

• Use of content standard (e.g., FGDC or ISO when
  replaces)
• Best used for semantics, domain specific
  terminology
• PREMIS
• Best used for management of resources over time
  using
• PREMIS Object entity
• PREMIS Event entity
• PREMIS Agent entity
• Useful to package resources metadata together
  to facilitate tracking of aggregation of
  resource(s), MD resource structure file
  inventory, e.g., METS

25
Issues Challenges

• What if domain specific MD is not available?
• If not, how can one get important info from data
  creators?
• How to determine what is truly necessary for use
  of data sets?
• Establishment of geospatial format registries
• Getting buy-in from geospatial domains for use of
  vocabularies, etc. (see Global Spatial Data
  Infrastructure http//www.gsdi.org/Default.asp )
  
• More research needs to be done on significant
  properties like that done by JISC DPC studies,
  e.g., SPs of vector images

26
Future directions for NGDA Project

• Further investigation of other geospatial formats
  including more vector based data such as
• layers of the National Atlas
• National Map (sections of California)
• Landsat 7 ETM imagery
• Derived data sets from Stanford faculty

27
Future directions, cont.

• Format Registry investigation - what should be
  included in a format registry for geospatial
• Contact with key vendors, e.g. ESRI,
  SafeSoftware, etc.
• Monitoring what others are doing with e-science
  social science data sets, e.g.,
• NCSU, Johns Hopkins
• National Australian Archive (NAA)
• JISC and DPC in the UK
• NDIIPP US Multi-state project
• Those using new DDI v 3.0 schema

28
References, contact info

• JISC DPC studies on significant properties
  http//www.dpconline.org/graphics/events/080407wor
  kshop.html
• See Duce and Nielsen papers
• Full paper available at http//www.ngda.org/resea
  rch.php
• National Geospatial Digital Archive
  http//www.ngda.org/index.php
• Examples of METS with PREMIS on METS public wiki
  
• http//www.socialtext.net/mim-2006/index.cgi?profi
  le_playground

29
Questions? / comments?

• Nancy J. Hoebelheinrich
• nhoebel_at_stanford.edu
• John Banning jwbanning_at_gmail.com