Semantic Grid Tools for Rural Policy Development & Appraisal Department of Computing Science,...
-
Upload
joseph-mcginnis -
Category
Documents
-
view
214 -
download
0
Transcript of Semantic Grid Tools for Rural Policy Development & Appraisal Department of Computing Science,...
Semantic Grid Tools for RuralPolicy Development & Appraisal
Department of Computing Science, University of Aberdeen
Department of Geography & Environment, University of Aberdeen
Macaulay Institute, Aberdeen
Outline
eSocial Science & The Grid The Semantic Grid PolicyGrid – Aims & Activities Supporting Social Simulation Metadata Challenges for eSocial Science Supporting Argumentation Summary
eSocial Science & The Grid
eScience UK DTI characterises as distributed global
collaborations enabled by the Internet. The concept of the Grid promises to provide access
to large data collections, near unlimited processing resources for running experiments and studies, and advanced visualisation facilities.
Grid Components Computational grid (Scavenging grid) Data grid
The Semantic Grid
Semantic Grid A vision of eScience infrastructure in which there is
much richer support for researchers to publish, share and re-use resources, integrate heterogeneous information, collaborate, access decision support tools, etc.
Central to this view is the integration of Grid technologies with Semantic Web technologies.
RDF Resource Description Framework OWL Web Ontology Language
The Semantic Grid
Data mining
Knowledge Discovery
Smart search
Social networking
Smart portals
Agents
Information Integration and aggregation
CourtesyCarole Goble,
University of Manchester
Ontologies
PolicyGrid
Aims To facilitate evidence-based rural, social, and land-use
policy-making through integrated analysis of mixed data types;
To demonstrate that Semantic Web/Grid solutions can be deployed to support various facets of evidence-based policy-making through the development of appropriate tools;
To focus on the authoring of relevant ontologies to support rural, social and land-use policy domains;
To investigate issues surrounding communication of semantic metadata to social scientists and policy practitioners;
To promote awareness of the Semantic Grid vision and supporting technologies amongst social scientists.
Builds upon work of the earlier Fearlus-Gpilot demonstrator project.
PolicyGrid
What are the methodological drivers behindour activities? A myriad of policy evaluation challenges facing
contemporary social scientists; Increased focus on methods and tools for integrated policy
evaluation; Increased emphasis on multi-method or mixed-methods
approaches to evaluation, where emphasis is placed on plural types and sources of data;
Diverse epistemological approaches and analytical techniques.
A key driver - evidence-based policy making – a mantra often summarised as meaning ‘what matters is what works’ (Cabinet Office, 1999).
Supporting Social Simulation
Fearlus Land-Use Model Case Study Aims
To serve a well-established simulationframework to the wider community
To support collaboration among socialscientists by providing a sharedco-laboratory environment forexperimentation.
Achievements Distributed simulation experiments run across Grid nodes. Simulation results annotated with metadata (RDF). Users can publish and share simulation model
parameters and re-run experiments. Support for creation of hypotheses, arguments. Ontology to support annotation of simulation resources.
Simulation Parameters
@beginenvironmentType Toroidal-MooreneighbourhoodRadius 1climateBSSize 0economyBSSize 16landParcelBSSize 0nLandUse 8pLandUseDontCare 0.0clumping NoneenvXSize 15envYSize 15nSubPops 2strategyChangeUnit 0.0neighbourNoiseMax 0.0neighbourNoiseMin 0.0breakEvenThreshold 8landParcelPrice 16subPopFile subPopDesc.sd
suddenchange150clim00000000000001000000000000000100000100000000000000111111011111111111111111111101111011111111111011111111011111
@beginNumberOfStrategyClasses: 3Class AboveThresholdProbabilityBelowThresholdNonImitativeProbabilityBelowThresholdImitativeProbability InitialProbabilityHabitStrategy 1.0 0.0 0.0 0.0RandomStrategy 0.0 1.0 0.0 1.0NoStrategy 0.0 0.0 1.0
Architecture
Desktop Application FEARLUS
Experiment Service
Upload Service
Repository Service
ELDAS Data Access Service
MODEL 0-6-5<CLASS>
FEARLUS Model<INTERFACE>
Model Factory<INTERFACE>
FEARLUS
META-DATA
My Workspace
Web Interface
Public Repository (Longwell)
Web Interface
My SQL
OGSA 3.2.1
WEB/GRID SERVICES FEARLUS MODEL INTERFACE
JDBC4ELDAS
My Workspace
Simulation Workflow Support
Taverna workflow tool
Allows scientists to describe and enact their experimental processes in a structured, repeatable and verifiable way.
MetaData Challenges for eSocial Science
Ontological Approach:– Universally shared conceptualisation of a domain of
discourse.– Provides a controlled vocabulary.
How to capture fuzzy/vague concepts?– sustainability, accessibility, poverty …
How to make different conceptualisations of a domain of discourse co-exist?– Differences in granularity.– Inconsistent points of view.– Meaning is often fluid, contextual.
There will never be just one ontology![In social science or any other activity]
Annotations - Semantic Web View
NVivo
Country Country
Political Office
City
Place
Annotation - assert facts usingterms (metadata in RDF).
Represent terms and theirrelationships (ontology in OWL).
Annotations help to connectWeb resources.
Annotations - Qualitative Social Science View
Qualitative data analysis tools such as NVivo.
Can we combine the Semantic Web view withthe qualitative analysis approach?
Folksonomies - A Solution for eSocial Science?
Ontologies are often seen as a “top-down” solution.– Will the social science community accept this?
Folksonomy– Derivation: “folk” + “taxonomy”– Collaboratively generated, open labelling system.– Social networks and collective intelligence.– Power derived from community “buy-in”.– Problem of meta-noise…
Folksonomies - A Solution for eSocial Science?
Folksonomies - A Solution for eSocial Science?
Supporting Argumentation
Arguments & Evidence
PolicyGrid Team
Project Investigators– John Farrington (Geography & Environment)– Gary Polhill, Nick Gotts (Macaulay Institute)– Pete Edwards, Alun Preece, Chris Mellish
(Computing Science)
Project Staff– Abdelkader Gouaich, Feikje Hielkema,
Edoardo Pignotti, ChuiChing Tan
www.policygrid.org