Long Term Preservation of Scientific Models using Semantic Web Services

1
Long Term Preservation of Scientific Models
(using Semantic Web Services)
Jane Hunter, jane_at_dstc.edu.au
2
PANIC Project

• Objectives
• Address the long term preservation and
  accessibility of (composite) digital objects
• Partners
• DSTC, UQ
• NLA, APSR

3
Objectives

• Provide an Integrated Preservation Framework
  which supports
• Large, heterogeneous, distributed collections
• Multiple formats, composite digital objects
• New emerging formats, software, recommendations
• New migration, emulation services
• Recommender services/decision support
• Flexible, Dynamic, Scalable, Extensible
• Sustainable - cost-effective, semi-automated
• Collaborative effort!

4
Existing Tools

• JHOVE, NLNZ, PREMINT Metadata tool
• OCLCs INFORM, Cornells VRC, SHERPA risk
  assessment -gt notification services
• GDFR, PRONOM, DCC-RIR Format registries
• VersionTracker, IIPC Software Registries
• XENA, TOM Conversion/migration services
• IBMs UVC (Universal Virtual Computer)
• Koninklijke Bibliothek - Emulation services

5
(No Transcript)
6
Steps

• Archival selection and capture of digital
  object(s) preservation metadata
• Risk assessment and notification of potential
  obsolescence
• New recommendations, formats, software versions
• Service Specification and Request
• Emulation or Migration
• Inputs/Outputs
• Cost
• Speed
• Remote/Distributed/Local
• Reliability
• Lossiness
• Select, Compose, Invoke Preservation Service
• Record preservation events

7
PANIC Architecture
MySQL databases
Sesame RDF Store
Apache AXIS
8
Notification component
Obsolescence detector periodically compares the
preservation metadata for each object with
registries to determine when object is at risk
of obsolescence
9
Invocation component

• Service Discovery provides a user interface so
  collections manager can specify the type of
  preservation service they are looking for.
• Service Selection presents the services
  retrieved by the Discovery agent for selection.
• Service Invocation invokes the chosen service
  and updates the preservation metadata where
  necessary

10
OWL-S Preservation Extensions
11
Discovery component

• Discovery Agent - matches service request against
  OWL-S descriptions of Preservation Web services
• Returns a ranked list of Preservation Web
  services that match the request

Sesame RDF Store
12
Provider component

• Provider Agent either
• retrieves and invokes preservation service
  locally or
• Invokes preservation service remotely

13
Hypothetical Example

• Russel Coight is an astronomer at the Australian
  Telescope National Facility (ATNF)
• Large collection of astronomy images in TIFF 5.0
  format
• ImageViewer 1.0 used to view TIFF images
• New version of ImageViewer (2.0) no longer
  supports TIFF 5.0
• RLG recommends that TIFF format be replaced by
  JPEG2000 for archival

14
(No Transcript)
15
(No Transcript)
16
(No Transcript)
17
(No Transcript)
18
(No Transcript)
19
(No Transcript)
20
(No Transcript)
21
eScience Workflow
Organization B
Organization A
Organization C

t1
t6
t7
t8
t5
t8

Initiate New Experiments
Statistical Analysis
Microscopic Images
Semantic Indexing
t3
t4
t2
MatLab Image Processing
Data Exploration Hypothesis Testing
Capture Results
Capture Spectrometry data
Conduct Experiments
BPEL4WS workflow based on web services
22
Scientific Discovery Process

• Inception of the idea
• Analysis of prior work
• Experimental design
• Capture the empirical/observational data
• Analyse, process, interpret and annotate the
  data
• Visualization of the data
• Formulation of an hypothesis, construction of
  conceptual and/or numerical models
• Verify and refine the model by comparison with
  new experimental data
• Document and publish the findings

23
Components

• Prior work - pre-existing data, models,
  hypotheses or publications
• Objective, hypothesis
• Experimental data - instrumental conditions,
  settings and parametric ranges or constraints,
  assumptions
• Data sets numerical data, survey/questionnaire
  data, images, video, audio, spectral data,
  real-time sensor data
• Formulae, rules, hypotheses, mathematical
  functions
• Conceptual models - axioms, models and metaphors
• Software
• source code, executables, applets or links to web
  services
• Hardware instruments and computers
• Visualizations 2D, 3D imagery, graphs, tables,
  charts, diagrams, animations
• Textual - publications, reports, documentation,
  annotations, bibliographies, reviews

24
Harmony ABC Model
25
Extended ABC Ontology
description
objectives
parameters
Experiment
conditions
Simulation
results
Processing
Numerical
Textual
Data
Image
Model
Graphical
Design
Audiovisual
Hypothesis
Mapping
Theory
26
Modelling Provenance
Experiment
Processing
Objectives
State3
State1
State2
Type
Type
Model
Input
Input
Event E1
Event E2
Output
Output
Visualization
Action
Action
Context
Context
Tool
Tool
Scope
MatLab
Role
Role
Date/ Time
Place
Agent
Conditions
Date/ Time
Place
Agent
Agents can be people or software e.g., web
services
27
Scenario
Tools to enable Construction Description Publishi
ng Of Scientific Model Packages
Private Workspace Project1 Project2 Project3
Shared Workspace Project1 Project3
RDF
SRB/MCAT
DSpace RDF DataStore
Distributed Databases

• Institutional Repository
• Scholarly Publications
• Scientific Model Packages
• eLearning Objects

28
Model to predict LE of CD-ROMs
RDF Package Title Creator Description Type
image_of
analysis_of
Average LE 1/T exp (A B/T)
derived_from
Each component Has software, OS,
hardware Dependencies interdependencies
refers_to
Slattery, O., Lu, R., Zheng, J., Byers, F., Tang,
X. "Stability Comparison of Recordable Optical
Discs- A study of error rates in harsh
conditions," Journal of Research of the NIST,
109, 517-524, 2004
29
Composite Objects

• Use RDF/XML to package metadata, component
  objects and relationships
• METS, MPEG-21 DIDL, XFDU, IMS-CP
• Maintain preservation metadata for
• Composite object
• Atomic components
• Maintain index of file formats
• Monitor atomic objects first
• JPEG -gt JPEG-2000
• PDF -gt PDG
• Then check currency of composite objects
• MPEG-21 DIDL V1 -gt MPEG-21 DIDL V2

30
Future Directions

• Ongoing refinement and evaluation
• within real archive/library (PANDORA)
• Integrate - GDFR, PRONOM, TOM, XENA, media
  longevity estimates, risk assessments
• Trusted services - quality ratings
• Grid Services - Web Services Resource Framework
  (WSRF)
• Composite services, composite objects

31
Reference

• http//metadata.net/panic
• contact jane_at_dstc.edu.au