ontologies for emergency & disaster management
DESCRIPTION
Ogc meeting march 2014 OGC OWS-10 Cross-Community Interoperability Ontologies for Emergency & Disaster Management (The application of geospatial linked data)TRANSCRIPT
Copyright © 2014 Image Matters LLC. All rights reserved. | www.imagemattersllc.com
OGC OWS-10
Cross-Community Interoperability
Ontologies for Emergency & Disaster Management(The application of geospatial linked data)
March 25, 2014
Stephane Fellah, Chief Knowledge Scientist
Copyright © 2014 Image Matters LLC. All rights reserved. | www.imagemattersllc.comPage 2
Outline
• Summary of our approach
• Core Geospatial Ontology
• Core Incident Ontology for E&DM
• Next steps
• Key takeaways
Page 2
Copyright © 2014 Image Matters LLC. All rights reserved. | www.imagemattersllc.comPage 3
Objectives of Our Research & Development
• Prove core concepts and state of readiness of
semantic-based approaches to interoperability
• Build, test and demonstrate Core Geospatial
Ontology
– Key new building block in achieving geospatial
interoperability
– Enabler for producing and sharing Geospatial Linked Data
(shared geospatial knowledge)
• Build, test and demonstrate core ontologies to enable
E&DM interoperability
• Implement a true semantic-enabled service that
demonstrates semantic integration and
interoperability across disparate geospatial sources
(a prototype “Semantic Gazetteer”)
Page 3
Copyright © 2014 Image Matters LLC. All rights reserved. | www.imagemattersllc.com
Data-to-Knowledge Integration Services:
“Crossing the Infocline”
Data-Centric World
(Today)
• Unsustainable cognitive load on
user to fuse, interpret and make
sense of data
• Interoperability is brittle, error-
prone and restricted due to lack of
formal semantics
• High cost of integration
Knowledge-Centric World
(Our Goal)
• Semantic-enabled services reduce
burden by “knowledge-assisting” user
• Semantic layer provides unambiguous
interpretations and uniformity… “last
rung in interoperability ladder”
• Agile, fast and low cost integration.
4
Copyright © 2014 Image Matters LLC. All rights reserved. | www.imagemattersllc.comPage 5Page 5
Value Proposition of Knowledge-Centric Approach (1)
Issues with Current Data-
Centric Approaches Knowledge-Centric Approach Increased Value
Data model standardization relies
upon homogeneous data
description and organization.
Employs a standards-based formal, sharable
framework that provides a conceptual domain
model to accommodate various business needs.
• Allows decentralized extensions of the domain
model
• Accommodates heterogeneous
implementations of the domain model (lessens
impact on systems; reduces cost)
• Shareable machine-processable model and business rules; reduces required code base
Increases the chance for multiple
interpretations and
misinterpretations of data.
Encodes data characteristics in ontology. • Increased software maintainability
• Improved data interpretation and utility• Actionable information for the decision maker
Data model implementations have
limited support for business rules,
and lack expressiveness.
Standards-based knowledge encoding (OWL,
SPARQL Rules) captures formal conceptual
models and business rules, providing explicit,
unambiguous meanings for use in automated
systems.
• Reduction of software and associated
development cost
• Conceptual models and rules that provide
enhanced meaning, thus reducing the burden
on users
• Unambiguous interpretation of domain model; greater consistency in use
Presumes a priori knowledge of
data utility. Semantics are pre-
wired into applications based
upon data verbosity, conditions
and constraints.
Encoding the conceptual model and rules
explicitly using OWL enables rapid integration of
new/changed data. Software accesses data
through the “knowledge layer” where it’s easier to
accommodate changes without rewriting software.
• Reduced software maintenance due to data
perturbations
• Software quickly adapts to evolving domain
model
• New information are readily introduced and understood in their broader domain context
Copyright © 2014 Image Matters LLC. All rights reserved. | www.imagemattersllc.comPage 6Page 6
Value Proposition of Knowledge-Centric Approach (2)
Issues with Current Data-Centric
Approaches Knowledge-Centric Approach Increased Value
Implementations are inflexible when data
requirements change. Whenever business
rules and semantic meaning are encoded in a
programming language, changes impact the
full development life cycle for software and
data..
Uses an ontology that contains a flexible,
versatile conceptual model that can
better accommodate the requirements of
each stakeholder in the business
domain.
• Increased flexibility to accommodate
stakeholder needs; Decentralized and
organic evolution of the domain model
• Changes only impact affected stakeholders,
not others; reduces software updates
• Software adapts to domain model as
ontology evolves
• The enterprise can better keep up with changing environment/requirements
Requires that data inferencing and validation
rules are encoded in software, or delegated
to human-intensive validation processes.
Uses a formal language (OWL) that
provides well-defined semantics in a
form compliant with off-the-shelf software
that automates data inferencing and
validation.
• Employs off-the-shelf software for
inferencing and validation
• Reduction of validation and testing in the
development process
• Uses all available data from sources,
including inferences, while accommodating cases of missing/incomplete information
Copyright © 2014 Image Matters LLC. All rights reserved. | www.imagemattersllc.comPage 7
Crossing
The
Infocline
Physical
Logical
Conceptual
Data & Analytic Services
Business Apps
Data-centric services impose
excessive cognitive load on
analysts
All Source All Source
Knowledge-assisted services
enhance triage, fusion, dot-
connecting, pattern detection,
inferencing
and sense making
Knowledge–assistedSemantic Services
Data-Centric Knowledge-Centric
A Paradigm Shift from
Data-Centric to Knowledge-Centric
Business Apps
Data & Analytic Services
Copyright © 2014 Image Matters LLC. All rights reserved. | www.imagemattersllc.comPage 8
What Linked Data Is About?
Tim Berners-Lee Vision: “… It’s not just about putting data on the web. It
is about making links, so that a person or machine can explore the web of
data. With linked data, when you have some of it, you can find other
related data.” By adding formal semantics and context to Linked Data,
it becomes “understandable” by software.
For the web to remain robust and grow, the following rules (standards) must apply:
• Use URIs as names for things
• Use HTTP URIs so that people can look up those names
• When someone looks up a URI, provide useful information, using the
standards (RDF, OWL, SPARQL)
• Include links to other URIs so that they can discover more things.
★ Available on the web
★★ Available as machine-readable structured data
★★★ Non-proprietary format
★★★★ Use open standards from W3C (RDF and SPARQL)
★★★★★ Link your data to other people’s data to provide context
5 ★ Rating for Linked Open Data
Why Linked Open Data?
Semantics and Context
Copyright © 2014 Image Matters LLC. All rights reserved. | www.imagemattersllc.comPage 9
Vision: Towards a Web of Shared Knowledge
Page 9
The train has already left the station…… an entire ecosystem of shared linked data exists
Copyright © 2014 Image Matters LLC. All rights reserved. | www.imagemattersllc.comPage 10
Summary of IM Contributions to OWS-10
• Geospatial ontologies
• Incident ontology
• Semantic service components
• Prototype “Semantic Gazetteer” service (GeoSPARQL-enabled)
that unifies access to 5 gazetteer sources – produces
“geospatial linked data” that can be shared across the Web
Page 10
Copyright © 2014 Image Matters LLC. All rights reserved. | www.imagemattersllc.comPage 11
Geospatial Ontologies Overview
• In-kind contribution from Image Matters to OGC community (8+ years of
development and testing)
• Core cross-domain geospatial ontologies
• Candidate foundation ontologies to bootstrap the Geospatial Semantic
Web
• Design criteria:
– Minimalist semantic commitment
– Modular
– Extensible
– Reusable
– Cross-domain
– Leverage existing standards
• Benefits– Multilingual support
– Linkable to other domains
– Sharable and machine-processable
– etc. (see slides 5 & 6)
Page 11
Copyright © 2014 Image Matters LLC. All rights reserved. | www.imagemattersllc.comPage 12
Core Geospatial Ontologies
Page 12
Mereology
Collections
Quality
Spatial Entities
Spatial RelationsSpatial Attributes
IdentifiersDatatypes
Upper ontology
Geometry
Measure
Reference Systems
Topology
EventTemporal EntitiesTemporal Relations
Temporal RSSRS
Quantity
Temporal Quantity
Spatial Quantity
Temporal
Role
Event Relations
Measurement Scale
Spatial
Measure
Common
Math Entities
Math RelationsMath Ops
Math
Event
Utilities
Based upon solid theoretical foundations
Copyright © 2014 Image Matters LLC. All rights reserved. | www.imagemattersllc.comPage 13
Semantic Gazetteer
Page 13
NGA Gazetteer
(Interactive Instrument)
Canadian Topo DB
(Compusult)
USGS Gazetteers
(Compusult)
WFS-G WFS-G WFS-G
GeoSPARQL Service
Semantic Mapping
ComponentSemantic Mapping
Component
Semantic Mapping
Component
Semantic Mapping
Component
Geonames PostGIS
(Image Matters)
Gazetteer Mappings
RDF Store
Client
(Pyxis)
Copyright © 2014 Image Matters LLC. All rights reserved. | www.imagemattersllc.comPage 14
Emergency Management Challenges
• Analysts and operators need to quickly triage, fuse,
connect dots, detect patterns, infer insights, and
make sense of the flow of incident information to get
an unambiguous Common Operational Picture
• Need to integrate and interpret incidents,
observations, mutual aid requests, alerts, etc.
generated across a multi-agency, multi-jurisdictional
spectrum
• Limited interoperability between agencies due to
different protocols, taxonomies, models and
representations (stovepipes are still there!)
Page 14
Copyright © 2014 Image Matters LLC. All rights reserved. | www.imagemattersllc.comPage 15
Incident Model Design Tenets
• Define a Core Incident Model applicable across domains, agencies and
jurisdictions
• Minimal semantic commitment (focus on core concepts and properties)
• Leverage Core Geospatial Ontologies
• Accommodate different Incident Model Profiles and Taxonomies
• Use Linked Data standards (RDF, SPARQL, RDFS, OWL, LDP)
– OWL used to model ontologies
– SKOS used to model taxonomies
• Sharable and machine-processable
• Linkable to other domains
• Multilingual support
Page 15
Copyright © 2014 Image Matters LLC. All rights reserved. | www.imagemattersllc.comPage 16
Incident Model
Page 16
Copyright © 2014 Image Matters LLC. All rights reserved. | www.imagemattersllc.comPage 17
Unified Semantic Incident Model
Page 17
NIEM
EXDL
NDEX
Unified
Incident model
NIEM
Converter
EDXL
Converter
NDEX
Converter
Other
Converter
NIEM
EDXL
NDEX
NIEM
View
EDXL
View
NDEX
View
Other
ViewOther Other
Profile1
Profile2
Profile3
Core
Incident
Model
Incident Data
Store
Data to knowledge Mapping
Copyright © 2014 Image Matters LLC. All rights reserved. | www.imagemattersllc.comPage 18
LEAPS Datasets
• Homeland Security Working Group
– Encoded taxonomies of Incident and Natural Events in SKOS
– Encoded symbols using Point Symbology ontology
• OpenFEMA disaster summaries
– Encoded disaster summary in linked data
• 911 Seattle Police dataset
– Encoded incident data as linked data
• Worldwide Incident Tracking System (WITS)
– Encoded Terrorist Incidents as linked data
• Abu Dhabi Police Model
Biggest challenge was the lack of good quality data.
Page 18
Copyright © 2014 Image Matters LLC. All rights reserved. | www.imagemattersllc.comPage 19
Homeland Security Work Group Symbology
Page 19
Copyright © 2014 Image Matters LLC. All rights reserved. | www.imagemattersllc.comPage 20
Proposed Roadmap: Next Steps
• Towards comprehensive Core Emergency Management Ontologies:
– Ontology for Organizations, covering governmental and non-governmental organizations involved in
E&DM; jurisdictions, roles and responsibilities of agencies and staff, type of resources they can provide
(based on W3C organization ontology)
– Ontology for Resource Management: Characterization of EM resources, including personnel and
equipment (see NIMS standard)
– Ontology for Response Activities: Covers aspects related to workflows of EM activities and services
such as medical services, aid delivery, food and shelter, victim triaging, on-site treatment, transportation
– Ontology for Communication: Covers communications (request, response, acknowledgment, alert, etc.)
and E&DM related message types (request/response for status, resources, information, deployment,
quote, requisitions, etc). (see EDXL-RM)
• Exercise the robustness of Core EM ontologies by developing profiles for different emergency-related
domains, agencies and jurisdictions
• Define architecture for Semantic Emergency Management System leveraging the Core EM ontologies
• Implement services to perform semantic mediation of incident information and representations
• Investigate, develop and test reasoning and inference for Incident/Resource/Communication Management
• Call for Datasets at International/Federal/State/Local levels
Page 20
Copyright © 2014 Image Matters LLC. All rights reserved. | www.imagemattersllc.comPage 21
Key Takeaways
• The core concepts espoused herein are solid and
repeatable
• Semantic-based interoperability can be achieved with
current technology
• A Core Geospatial Ontology is foundational to
sharing geospatial data and knowledge
• Core E&DM ontologies are crucial to interoperability
• Semantic Gazetteers, and many other such services,
illustrate the power and value of semantic-based
interoperability and services
– Can be readily added to existing “data-centric” infrastructure
Page 21
Copyright © 2014 Image Matters LLC. All rights reserved. | www.imagemattersllc.com
Questions?
Contact Information
Stephane Fellah
Chief Knowledge Scientist
Image Matters LLC
Leesburg, VA
USA
+(703) 669 5510
Page 22
Copyright © 2014 Image Matters LLC. All rights reserved. | www.imagemattersllc.comPage 23Page 23
Homeland Security Work Group Symbology