GEON Cyberinfrastructure Workshop

1
Workflow-Driven Ontologies for the Geosciences

• Leonardo Salayandía
• The University of Texas at El Paso

2
Overview

• Background
• Cyberinfrastructure
• Ontologies
• Workflows
• Purpose of this talk
• The Workflow-Driven Ontology approach
• Knowledge capture
• Workflow creation from WDOs
• Benefits of WDOs
• Status
• Summary

3
Cyberinfrastructure
S-wave tomography models
GPS plate motion vectors
GEON IDV (Integrated Data Viewer)
Global Strain Rate Map
http//geon.unavco.org
4
Cyberinfrastructure
S-wave tomography models
GPS plate motion vectors
GEON IDV
Global Strain Rate Map
Distributed tools and applications
Distributed sources of information Information in
different formats
5
Cyberinfrastructure

• People and resources connected through the web
• Enhanced collaboration over distance, time, and
  disciplines
• Interoperate across institutions and disciplines
• Preserve and maintain availability of software
  and data

6
Cyberinfrastructure

• People and resources connected through the web
• Enhanced collaboration over distance, time, and
  disciplines
• Interoperate across institutions and disciplines
• Preserve and maintain availability of software
  and data

7
Ontologies

• A specification of a conceptualization
• Concepts (or classes of objects)
• Concept1 S-wave tomography model (TM)
• Concept2 Geospatial representation
• Relationships between concepts
• S-wave TM HAS Geospatial Representation

8
Workflows

• Recipes for accomplishing some complex task
• Composition of service modules (CI services)
• Automate tedious and time-consuming tasks
• Useful for experiment replication
• Example

9
Workflows

• Recipes for accomplishing some complex task
• Composition of service modules (CI services)
• Automate tedious and time-consuming tasks
• Useful for experiment recreation
• Example

S-wave tomography data
Create Model
S-wave tomography model
Service to get the data
Service to transform data
Transformed data outcome
10
Cyberinfrastructure
B. Ludäescher, 2006
11
Cyberinfrastructure
Ontologies
Workflows
B. Ludäescher, 2006
12
Purpose of talk

• Show an approach for scientists to capture
  knowledge in a way that can be leveraged towards
  CI
• Create ontology specifications
• Generate workflows from ontologies

13
Purpose of talk

• Show an approach for scientists to capture
  knowledge in a way that can be leveraged towards
  CI
• Create ontology specifications
• Generate workflows from ontologies

Workflow-Driven Ontologies (WDOs)
14
Example Gravity WDO
I use geophysical data to elucidate the tectonic
development of the North American craton
Get the data
Gravity Data
Dr. Randy Keller
I want to produce a gravity data contour map.
These are the steps that I go through to do it
Create a grid of uniformly distributed points
from this data
Grid
Geoscientist
Use the grid as input to render the map
Contour Map
15
Capture Knowledge
Gravity Data
Grid
Different types of Information
Contour Map
16
Capture Knowledge
Information
Gravity Data
Raw Data
Is converted to
Grid
Processed Data
How is the information transformed?
Is rendered into
Contour Map
Product
17
Capture Knowledge
Information
Methods
Gravity Data
Is input into
Raw Data
Gridding Algorithm
Is converted to
Outputs
Grid
Processed Data
Is input into
Contouring Algorithm
Is rendered into
Contour Map
Outputs
Product
18
Class Hierarchy for WDOs
Root
Information
Methods
Data
Product
Processed Data
Raw Data
Gridding
Contouring
Gravity Data
Grid
Contour Map
Common classes for all WDOs
Classes specific to the Gravity WDO
19
Workflow specification generated from Gravity WDO
Root
Mapping between WDO classes and CI services
Information
Methods
CI Service2 Gridding
Data
Product
Processed Data
Raw Data
Gridding
Gravity Data
Grid
Outputs
Is input into
CI Service1 Gravity Data Extraction
Result
20
From workflow specification to workflow
implementation

• Workflow engines
• Kepler scientific workflows (GEON et al.)
• OWL-S (Semantic Web)
• Many others
• Workflow specifications produced from WDOs can
  potentially be realized in any service-oriented
  workflow engine

21
Benefits of WDOs

• Scientific products drive the creation of the WDO
• Incremental development
• WDO serves as roadmap for future CI service
  development
• Identify missing services for potentially useful
  workflows
• Generated workflows serve as a gauge for the
  usefulness of an ontology

22
Status

• Gravity WDO prototype
• Workflows in the process of being implemented in
  the Kepler Scientific Workflow Engine
• WDO Assistant and API software

23
The Gravity WDO

• First WDO prototype (Flor Salcedo, Randy Keller,
  and Ann Gates)

24
Status

• Gravity WDO prototype
• Workflows in the process of being implemented in
  the Kepler Scientific Workflow Engine
• WDO Assistant and API software

25
WDO Assistant and API

• Prototype built on top of the Jena API
• Java programming language
• Three modes of operation
• Brainstorming
• Elicitation
• Workflow Generation

26
WDO Assistant and API

• Brainstorming mode
• Scientists define concepts that relate to CI
  information and methods

27
WDO Assistant and API

• Elicitation mode
• Scientists define relationships between concepts

28
WDO Assistant and API

• Workflow Generation mode
• Scientists choose information concept for which
  to generate a workflow, as well as target
  workflow engine

29
Future Work

• CI-Miner
• Provenance information
• Trust information
• Preferences

30
CI Miner
CI Background Tools
Legend
ontologies
Knowledge capture
calls
OWL onts.
JENA
Protégé, SWOOP
uses
WDOs
creates
WDO Assistant
WDO API
CI-Base (IWBase)
Generic CI Portal
WFGen
Service execution
Atomic OWL-S Service
PSW
Composite OWL-S Service
A Service
OWL-S API
Answer/ provenance visualization
CI-Browser
PML
Trust Recommendation
TrustNet
CI-Trust
CI-Browser
31
Summary

• In order to realize the goals of CI there is a
  need to
• Capture domain knowledge
• Use the domain knowledge to glue resources
  together
• The WDO approach
• Allows scientists (not computer programmers) to
  incrementally capture knowledge as needed
• Facilitates communication between scientists and
  computer programmers to produce CI resources that
  stick to other resources

32
Thank you