Transcript Slide 1

GeoVISTA Center,
Department of Geography,
Pennsylvania State University
Mark Gahegan
Bill Pike
Sachin Oswal
Gary Sheppard
Gary Liu
Brandi Nagle
Junyan Luo
Sharing Our Understanding
Of Earth Science Resources
A knowledge management portal to
support collaborative geoscience
CYBERINFRASTRUCTURE FOR THE GEOSCIENCES
www.geongrid.org
Introduction, motivation & year 2 goal
 Making electronic geoscience resources more
available is not enough…
 We need to be able to describe these resources more
effectively…
 To be successful, contributing and finding
resources must become an integral part of the
way scientists/educators work
 Major goal for year 2…Develop visually-based
tools to help geoscientists organize, describe,
and gain access to the GEON resources
CYBERINFRASTRUCTURE FOR THE GEOSCIENCES
www.geongrid.org
Knowledge management for collaborative geoscience
Representation
• top-down ontology languages
• bottom-up context, situations (provenance)
• visual appearance, signification
• history & evolution
• alternative descriptions
instantiation
Capture
conceptualization

• collaborative web interface
• diagramming tools
• text mining tools
• importing existing ontologies
• workflow discovery
Usage
• ontology mediation services
• ontology similarity measures
• browsing conceptual structures
• shared virtual workspace
CYBERINFRASTRUCTURE FOR THE GEOSCIENCES
www.geongrid.org
Representation




Ontology languages (OWL, RDF, DAML+OIL)
Association histories of how resources are used
Visual appearance / signification serialization
Additional descriptive information / resources
</owl:Class>
<owl:Class rdf:ID="Marsh">
<rdfs:subClassOf rdf:resource="#CoastalRegion"/>
<rdfs:subClassOf rdf:resource="#WetlandRegion"/>
</owl:Class>
…
Fragment of OWL ontology from NASA’s EarthRealm project
CYBERINFRASTRUCTURE FOR THE GEOSCIENCES
www.geongrid.org
Contextualizing science
“In science, numerous lines of investigation interweave to
delineate a type of rationality that is historically situated
and practical, and involves choice, deliberation, and
judgment.”
Richard Bernstein
Beyond Objectivism and Relativism: Science, Hermeneutics, and Praxis
- Richard Bernstein
Beyond Objectivism and Relativism: Science, Hermeneutics, and Praxis
Our aim is to contextualize resources through experiences; this is
crucial for understanding in domains that are highly interpretive
Put another way, what do feeding ducks have in common with
50% of our understanding?
CYBERINFRASTRUCTURE FOR THE GEOSCIENCES
www.geongrid.org
Three problems with a solely ontological approach
 Top down knowledge (ontologies) only get you so far… other
kinds of (bottom up) knowledge are also very important &
useful
 Use-cases (situations surrounding the use of resources)
 Social networks
 Most current ontologies are static resources…
 Our understanding is dynamic & continually evolving
 Unless ontologies are community-owned, dynamic resources they will
soon become part of the problem, not part of the solution
 What happens to all the thousands of resources that predate
ontologies?
 The cost of retro-fitting ontologies is prohibitive.
CYBERINFRASTRUCTURE FOR THE GEOSCIENCES
www.geongrid.org
Associations
CYBERINFRASTRUCTURE FOR THE GEOSCIENCES
www.geongrid.org
Amazon Web Services, degrees of separation using the Amazing (Kevin)
Baconizer (www.baconizer.com)
From "How Maps Work" by MacEachren, Alan to "Oops I Did It Again" by Spears
Britney: 12 hops –
People who bought:
How Maps Work: Representation, Visualization, and Design - By Alan M. MacEachren
also bought:
Web Cartography - By M-J Kraak and Allan Brown
People who bought this also bought:
Seeing Through Maps: The Power of Images to Shape Our World View - By Ward Kaiser and Denis Wood
Mapping: An Illustrated Guide to Graphic Navigational Systems - By Robert Fawcett-Tan
What is a Designer: Things, Places, Messages - By N Potter and R Kinross
Reinventing the Wheel - By Jessica Helfand
Photobooth - By Babbette Hines
MTV Photobooth - By MTV and Rizzoli International Publications
Stages - By Britney Spears and Sheryl Berk
Britney Spears - By Britney Spears
Baby One More Time (+5 Bonus Tracks) - By Britney Spears
Oops I Did It Again - By Britney Spears
CYBERINFRASTRUCTURE FOR THE GEOSCIENCES
www.geongrid.org
Capturing use-cases
Who created that concept / resource?
When was it created?
Has it been modified recently?
Who has used it? …
What did they do with it?
Such questions add a rich context by
capturing situations surrounding resource
usage
CYBERINFRASTRUCTURE FOR THE GEOSCIENCES
www.geongrid.org
Resource usage data
logged usage data (Oracle, MySQL)
CYBERINFRASTRUCTURE FOR THE GEOSCIENCES
www.geongrid.org
Mining association rules from use-case logs
 Association rules are mined from user action logs (uses the
WEKA (Waikato Environment for Knowledge Analysis) API
that implemented the Apriori algorithm (Agrawal, R. and
Srikant, R., 1994).
 Tools added for data preprocessing and classifying:
 attribute selector: allows user to select a subset of data attributes.
 data filters: allows user to define filters to convert String, Time,
Numeric data in any attribute column to nominal data for association
mining.
CYBERINFRASTRUCTURE FOR THE GEOSCIENCES
www.geongrid.org
Data mining tools (association rules)
Results &
sensitivity
settings
Data Filter - String
Attribute Selector
Design
Data Filter - Time
Data Filter - Numeric
CYBERINFRASTRUCTURE FOR THE GEOSCIENCES
www.geongrid.org
Capture: concept creation & harvesting (Codex, e-Delphi)
CYBERINFRASTRUCTURE FOR THE GEOSCIENCES
www.geongrid.org
Capture example
(Randy Keller’s gravity map from previous GEON
meeting)
CYBERINFRASTRUCTURE FOR THE GEOSCIENCES
www.geongrid.org
Supplemental material: e.g. educational resources
CYBERINFRASTRUCTURE FOR THE GEOSCIENCES
www.geongrid.org
Supplemental material: e.g. Google search results
CYBERINFRASTRUCTURE FOR THE GEOSCIENCES
www.geongrid.org
Google search (Google search API is built into Codex)
CYBERINFRASTRUCTURE FOR THE GEOSCIENCES
www.geongrid.org
Usage
codex demonstration
CYBERINFRASTRUCTURE FOR THE GEOSCIENCES
www.geongrid.org
CYBERINFRASTRUCTURE FOR THE GEOSCIENCES
www.geongrid.org
CYBERINFRASTRUCTURE FOR THE GEOSCIENCES
www.geongrid.org
CYBERINFRASTRUCTURE FOR THE GEOSCIENCES
www.geongrid.org
CYBERINFRASTRUCTURE FOR THE GEOSCIENCES
www.geongrid.org
CYBERINFRASTRUCTURE FOR THE GEOSCIENCES
www.geongrid.org
Managing groups & user workspaces
CYBERINFRASTRUCTURE FOR THE GEOSCIENCES
www.geongrid.org
Reusable knowledge structures afford…
 Private and shared knowledge spaces for describing
resources
 Provenance information produces a web of relationships
between resources
 Evolution and emergence of ideas within a community
 Discovery of points of agreement and divergence in
concept construction or problem-solving approaches
http://flatbox.geog.psu.edu/codex
CYBERINFRASTRUCTURE FOR THE GEOSCIENCES
www.geongrid.org
Example: questions you can ask
Gravitational anomaly dataset A








Is described by these concept map(s) / ontologies:
Was created in this way:
Plays a role in these workflow(s):
Has been used to fulfill these task(s):
Has been used by these people:
Is most often used with these method(s)
Has received the following review(s) / feedback:
Is similar to, or differs from, anomaly dataset B in the following
way(s):
CYBERINFRASTRUCTURE FOR THE GEOSCIENCES
www.geongrid.org
Future plans
 Add more perspectives onto resources into Codex (e.g.
working with Digital Library for Earth Science Education
(DLESE))
 Improve transition from one perspective to another
 Peer-to-peer implementation
 Improve transition between semi-formal concept maps
(provided by domain scientists) and formal (computable)
ontologies that are defined more rigorously.
 Experiment with Codex used live to capture conceptual
understanding (face to face and over the Web)
CYBERINFRASTRUCTURE FOR THE GEOSCIENCES
www.geongrid.org
Summary: projects we are perusing for GEON
1. Concept map / ontology visualization & management tools (ConceptVista &
Codex): searching & browsing of knowledge domains, and other resources.
2. Concept capture software (e-Delphi, Codex): developing vocabularies by
which resources and learning activities are described
3. Concept map / ontology versioning and comparison (differencing)
4. Concept uncertainty (fuzzy-rough set approach)
5. Use-Case Tools: logging and data mining (association rules)
6. Visualization and analysis tools: e.g. animated maps, scatterplots, 3D scenes,
cluster analysis, machine learning methods
7. Component assembly and deployment (GeoVISTA Studio): could help in
selecting and packaging activities into self-contained, deployable units.
8. Managing learning activities: Learning Activity Toolkit (Southampton, UK &
PSU)
9. Integration of concept management with DLESE API & strand maps
CYBERINFRASTRUCTURE FOR THE GEOSCIENCES
www.geongrid.org
Publications
1.
Pike W., Gahegan M, 2003, “Constructing semantically scalable cognitive spaces”, in: Spatial Information
Theory: Foundations of Geographic Information Science. Lecture Notes in Computer Science 2825, Kuhn W,
Worboys M, and Timpf S (Eds.). Springer-Verlag, Berlin: 332-348.
2.
MacEachren A M, Gahegan M, Pike W, 2004, “Geovisualization for constructing and sharing concepts”,
Proceedings of the National Academy of Sciences, Vol. 101.
3.
Gahegan M, Pike W, Ahlqvist O, Neff R, Yu C, “How much do we agree? A knowledge management system
to help represent and mediate concepts developed by collaborating human-environment researchers”
submitted to Annals of the Association of American Geographers.
4.
Gahegan, (2004). “Beyond tools: visual support for the entire process of GIScience.“ In: Exploring
Visualization (Eds. Dykes, J., MacEachren, M. and Kraak, J.-M.)
5.
Brodaric, B. and Gahegan, M. (in press) “Representing Geoscientific Knowledge in Cyberinfrastructure:
challenges, approaches and implementations”. GSA Special Papers volume.
6.
O’Brien, J. and Gahegan, M. (2004). “A knowledge framework for representing, manipulating and reasoning
with geographic semantics.” International Conference on Spatial Data Handling, Leicester.
7.
Gahegan, M. (2004).“The Future of GIScience? GRID Computing and the Semantic Web”. Keynote
address, GISRUK Conference, www.gisruk.org
8.
Pike W,Yarnal B, MacEachren A, Gahegan M,Yu C, (in press) “Infrastructure for collaboration: Building the
future for local environmental change”, to appear in Environment.
9.
Pike W. A., Ahlqvist O., Gahegan M., Oswal S., “Capturing context in collaborative science: Supporting
collaborative science through a knowledge and data management portal,” Workshop on Semantic Web
Technologies for Searching and Retrieving Scientific Data, at Second International Semantic Web Conference,
Sanibel Island, FL, October 2003.
CYBERINFRASTRUCTURE FOR THE GEOSCIENCES
www.geongrid.org
end
Questions?
CYBERINFRASTRUCTURE FOR THE GEOSCIENCES
www.geongrid.org
Supplemental slides
CYBERINFRASTRUCTURE FOR THE GEOSCIENCES
www.geongrid.org
Managing and sharing visual appearance
Concept
Hierarchies
A Hierarchical View
of the Concepts
Concepts are listed
Alphabetically
Currently We
Support RDF, OWL,
and XML.
Concept Graph
Style Editor
Styles describe how
concepts should be
rendered.
Different concepts
can have different
styles using property
filters
Styles can be
serialized using
XML-based StyledLayered Descriptor
Language, (SLD)
Concepts are Represented as Nodes, and their relations
are represented as Edges.
CYBERINFRASTRUCTURE FOR THE GEOSCIENCES
www.geongrid.org
Cyber-Infrastructure: underlying technologies
 Peer-to-Peer (P2P) Computing, software technology that enables
networked computers to communicate (exchange information) without a
common operating environment.
 The Information Power Grid (IPG) and Globus provide protocols
 Web Services, provide standards to describe, find & access remote resources.
 Web services mechanisms are integrated into the Grid model through the Open
Grid Services Architecture (OGSA).
 Semantic Web, describing and searching for web content using formalized
semantics (controlled vocabularies, taxonomies, ontologies)
 … as opposed to the current ‘chaos’, largely based on literals, popularity &
corporate sponsorship!
 Collaborative Knowledge Environments,
 Data & Knowledge portals
 Asynchronous discussions
 Video conferencing
CYBERINFRASTRUCTURE FOR THE GEOSCIENCES
www.geongrid.org
Towards a knowledge collaboratory
CYBERINFRASTRUCTURE FOR THE GEOSCIENCES
www.geongrid.org
gesopatial
infrastructure
An Integrated Approach to Distributed GeoCollaboration
infrastructure for geoscience
GEON
infrastructure for e-government/e-society
National
Map
HERO
Digital
Earth
infrastructure for homeland security
Geospatial
One-Stop
NGA:
NSGI
application
domains
research advances (in gray)
leveraged to meet challenges (in blue)
o
Ontology
and concept
browsers
Existing
metadata
standards
Semantic
web
integrating
knowledge
acquiring
knowledge
Ontology
creation
Browsing &
querying
knowledge
Knowledge
Infrastructure
constructing &
accessing
knowledge
Automated
indexing
tools
Supporting
knowledge
evolution
Ontology
mining /
harvesting
Representing
and sharing
perspectives
Semantic
indexing
Semantic search
Geospatial data
repositories
Meta-search
(ensemble
techniques)
Supporting
knowledge
communities
applying
knowledge
Collaboratories
Collaborative
visualization
Visually
mediating
understanding
Off-loading
ideas
Enabling
negotiation
Group Work
with geospatial
information &
technologies
Making decisions
Supporting
work
Distributing access practices
to knowledge
Dialogueenabled
interface
Content-object
replication kit
(CORK)
e-Delphi, ConceptVISTA,
& argument visualization
e
Geo /
Environmental
science
advancing science
K-12 science
& professional
development
Public/civic
planning/resource
management
enhancing prosperity & civil society
CYBERINFRASTRUCTURE FOR THE GEOSCIENCES
Emergency
response &
recovery
Strategic
threat
assessment
supporting homeland security
www.geongrid.org
Contexts:
Creation
Application
Represented by
Who did it?
Who should use it?
Collections of people
Where was it made?
Where does it apply?
Collections of sites /
scales
When was it made?
When does it apply?
Collections of temporal
intervals
How was it made?
How should it be used?
Collections of methods
and data
Why was it made?
Why should it be used?
Collections of research
questions, motivations,
theories
CYBERINFRASTRUCTURE FOR THE GEOSCIENCES
www.geongrid.org