semantic interoperability1 semantic interoperability net centric perspective presented to sicop team...

Semantic Interoperability 1

Semantic InteroperabilityNet Centric Perspective

Presented to SICoP Team

John A. Yanosy Jr.Chair NCOIC SII-WG

August 15, 2006


“Many of the problems we have identified can be categorized as “information gaps” – or at least problems with information-related implications, or failures to actdecisively because information was sketchy at best. Better information would have been an optimal weapon against Katrina. Information sent to the right people at the right place at the right time. Information moved within agencies, across departments, and between jurisdictions of government as well. Seamlessly. Securely. Efficiently.”

The Final Report of the Select Bipartisan Committee to Investigate the Preparation for and Response to

Hurricane Katrina


Semantic Interoperability Overview


Networking Tension

Independent Systems

Increased Networking

Network Benefits

Operational Costs

Low Costs

High Costs

- Increasing Complexity Costs

+ Improved & Faster Decisions

Low Network Benefits

High Network Benefits

A - Net Low Autonomic


Semantic Interoperability Context

System System

World

People People

1 H-H2 H-S3 H-W4 S-S5 S-W

1

22

3 3

4

Hypothesis- Semantic errors due to mutual misinterpretation cause unintended consequences in system interactions

5 5

Eachhas potentialfor semantic breakdown


Goals

Semantic Interoperability Intersection

KnowledgeInformation

Context

Actions/Services

Communications


Conceptual Framework of Frameworks

SemanticCommunication Framework (SOAP)

SemanticServices Framework (OWL-S)

SemanticCollaboration Framework (FIPA, ACL)

Semantic InformationFramework

(OWL,NIEM,DRM

C2IEDM)

SemanticContext Framework (?)

SemanticPolicy Framework (SWRL?)

Semantic Content

Framework(MPEG)

SemanticAutonomic Management (?)


Semantic Frameworks• Semantic Net Centric Interoperability Framework – architectural gereral layers and elements

enabling mutually consistent interpretation of interactions• Semantic Information Framework - layered model defining hierarchical semantic knowledge

and structures• Semantic Context Framework – situational context model• Semantic SOA Framework - layered model web services including Service Discovery Profiles

Semantic Metadata, Semantic Service Descriptions, Taxonomic Metadata enabling linking of service dependency relationships, metadata linking service descriptions to domain specific semantic data models, common services supporting capabilities required by all services

• Command and Control Semantic Adaptive Policy Control Framework - semantic framework enabling overall policy constraints on the Semantic Information Framework and the Semantic SOA Framework - provides command and control across all layers representing unique constraints by various COIs

• Semantic Autonomic Management Framework- provides semantic models enabling self management supporting operations (Self Configuration, Self Healing, Self Optimization, Self Security)

• Semantic Collaboration Framework - Intelligent Agent based Framework that can support adaptive mediation between all of the other frameworks and that provides dynamic collaborative formation of agents

• Semantic Communication Framework - framework encompassing support for metadata descriptions of message based communications, metadata representation of message content, models relating data elements of messages to Semantic Information Framework semantic data models, models defining semantic intention of message

• Semantic Media and Content Framework - framework enabling the semantic representation of the nature of the media content type, music, voice, image, etc. in such a way that adaptation can be provided to modify content for end to end and device adapation purposes - many meatdata standards already exist


Semantic Interoperability Issues


Semantic Incompatibility Issues

• COIs, Social, Organizational, Cultural Assumptions and Policies

• Domain Knowledge• Ontology Relationships and Harmonization• Logic(s) DL, FOL, Intensional, SWRL• Context Dependency• Semantic Expressibility, • Semantic Web – URI Networking, Discovery • Implementation Technology


Varying Semantic Representations

Signals

Explicit

Implicit

Data

Objects

Metadata

Domain Ontology

Taxonomies

Semantic Web

Syntax Structure

Domain Vocabulary, Schemas

Organized Hierarchical Classifications

Untyped Data

Semantic Knowledge Model & Logic

Relevant, Discoverable, Understandable Semantic Knowledge

Current Systems EmergingSystems

EmergingNetworks

OWL

OWL-S

XMLUMLMIF

RDFS

RDFXMLschemaWSDL

DoDTaxonomyUDDIDOM

UDDI, Domain,OntologiesContext Ontologies,CognitiveAgents,Mediation

NCOService& DataTenets

Context, Upper Ontology

C-OWLCOINIEEE SUMOIFFUpper CYCSWRL

Link 16ASCII

Dublin CoreDDMSTMLC2IEDMebXMLWordnet

Net Centric

Semantic Interoperability Across Ontologies

CYCFoaF

X


Semantic Interoperability Problems• Shared Knowledge - Semantic interpretation of shared information between systems

and also between systems and people is interpreted by humans in a collaborative manner when designing and developing systems, but typically the results of these collaborative semantic interpretations are not explicitly represented in the solution, rather they are implicit in the solution. This results in possible semantic misinterpretations for different system implementations that are supposed to have a common semantic interpretation of shared information.

• Situational Context – Without understanding context of a particular user’s situation, the user bears the burden and complexities of discovering and selecting appropriate system capabilities and desired information. In contrast context knowledge defining the relevant information required for a specific situation and perspective can be used to personalize a system’s response more appropriate to the user in a current situation. It also defines the situation and the domain knowledge important to it. Context theory and context aware applications are being developed to enable adaptation of system behavior to a participant’s context.

• Closed Semantic Network - assumes implicit semantics achieved through human interpreted specifications and related design activities. Systems within closed environments interoperate reasonably well as long as the operating environment, the expected use of the systems, and the system definitions themselves are consistent over time with little change. If any of these conditions are modified than the original semantic interpretations about system functionality, information exchanged, and expected behaviors have to be reevaluated.

• Open Semantic Network – allows for heterogeneous semantic environment with capability to extend additional semantic definitions. Problems of ontology harmonization, varying levels of expressibility, different models, different intentions and context


Semantic Interoperability Errors Occur Everywhere

• Client –web services interactions (Client interprets <SellStock> as post offer to sell, while web service interprets as sell at any price.)

• App – App operations and data exchanges (AppX interprets <Stock> as symbol of stock, while App Y interprets <Stock> as CurrPrice)

• Use of API interfaces (API labels passed argument for operation as “Stock”, with no semantic definition, object implementing API interprets “Stock” as quantity of items available in inventory.)

• Interpretation of Network protocols by network elements (Each NE interprets the protocol message according to the role it has in a Closed World Network and a shared protocol specification – Semantic intent of messages across Closed World Networks require reinterpretation in gateways)

• App interpretation of database information (Subtle misinterpretation of database meaning by new App results in inconsistent database state due to inappropriate updates by new app e.g., HealthcareProvider updates PatientStatus due to diagnostics, while FinanceAdministration updates PatientStatus due to InsuranceConstraints. In this case PatientStatus was originally used for health status, not Insurance status.

• Web services search – UDDI service profiles have no associated schemas or ontologies, resulting in semantic misinterpretation of keyword searches for services

• Information search – Without taxonomies of knowledge domain profiles searches will rely on data mining algorithms with too many non-relevant results

• Information integration and merging across apps, systems and databases – Biggest problem of semantic interoperability since multiple specifications and enterprise purposes are involved, as well different syntax information structures and constraints.

• Exchanged XML documents – only contains simple or complex data element definitions, no relationships between data elements or constraints about when data element instances can be created

• WSDL web service specifications - no semantics associated with WSDL service definitions, such that applications would have to be written to each WSDl service vocabulary, even when in the same application domain.


Closed Solutions(Implicit Semantics)

• Closed solutions are characterized by static aspects with implicit semantic interoperation between like systems due to:– explicit semantics defined in the requirements and

design stage, – implementations having weak traceability to

requirements and design semantics

• Results in brittle and complex semantic interoperability specifications not easily modifiable for interoperation with other systems, nor easily evolvable with changing requirements


Open Solutions(Explicit Semantics)

• Open solutions are characterized by dynamic aspects that enable explicit semantic interoperation at multiple levels of interaction between different systems due to:– explicit semantics defined and accessible in all phases– sharing of intensional semantic knowledge about context, intentions,

actions, capabilities, commitments and environment– simple universal communications speech acts enabling collaboration

between systems– separation of semantic concerns and explicit representations of

knowledge and system actions or services– ability to extend the universe of explicit knowledge used by systems

as new requirements and capabilities are desired– ability to discover, access, and share explicit knowledge in multiple

domains (context, capabilities, environment perspective, commitments, …)

– ability to dynamically marshal resources to broker semantics• Results in extensible and robust interoperability solutions resulting

from dynamic integration of disparate systems within a common semantic interoperability framework


A Universal Semantic Interoperability Framework


Universal Semantic Interoperability Model

Goals

Knowledge

Speech Acts

Context

Goals

Knowledge

Speech Acts

Context

PurposefulCommunications

Intentions,Services

Intentions,Services

Reasoning

Reasoning

Reasoning

Reasoning

Request, Committment

Shared Domain Knowledge

Perspective, Situation

Collaboration, Role

REACTIVE

COGNITIVE

ENVIRONMENT

Perception Perception

WorldModifyingActions


ENVIRONMENT

ENVIRONMENT

ENVIRONMENT


Transformation from Implicit to Explicit Semantic Interoperability

Full Semantic interoperability is enabled by embedding and sharing explict semantic representations of agent, system and environment goals, context, intentions, actions, available services, domain knowledge, and speech acts


Semantic Interoperibility Model

• SIOPM = <SM, WFF>, set of semantic models and well formed expressions entailed by each model

• SM = <SM1, …,SMn>, set of semantic models used by agents 1, …, n

• SM = <D, G, V, I, L, A, wff>, semantic model tupleD = Domain and individuals in domainG = Grammar defining syntax of well formed expressions, wffV = domain vocabulary for domainI = Interpretation function mapping domain vocabulary terms to

domain individualsL = Logic defining rules of reference and entailment for wffA = Axioms predefined in modelSM |= wff , wff entailed by Model M, ( |= Entailment operator)


Semantic Interoperibility Model• Mutual Semantic Entailment Between

Pairs of Agents Ai and Aj– Mi Mj – Mi Mj |= wff Mutual Semantic Entailment

• Non-Mutual Semantic Entailment Between Pairs of Agents Ai and Aj– Mi Mj

• (Mi|= wff) (Mj|= wff)

– Mi Mj • Mi Mj | wff


Goals

Semantic Interoperability Intersection

KnowledgeInformation

Context

Actions/Services

Communications


Semantic Interoperability Principles• Interoperability between systems and agents is purposeful and

informed by goals, contexts, and shared semantic domain knowledge models (whether explicit or implied).– actual world modifications are achieved through intentional

actions . – sharing of semantic environment knowledge provides a ‘situated

real world’ perception to enable better decisions about what actions or services are required to achieve goals (mapping of sensed data to perception concepts)

• Goals guide selection of intentions and execution of actions• An extensible network of semantic services with explicit semantic

representations enables interoperability independent of platforms and technology implementations, and provides a foundation for intentional actions within a purposeful, cognitive interoperable framework

• Communications occurs within few universal intentional categories (Speech Acts – request knowledge, commit to action, request action, … )

• Context defines relevant domain knowledge for a specific situation

• Useful Knowledge is organized in semantic domain models


Cognitive Semantic Interoperability Model

Goals

Knowledge

Speech Acts

Context

Goals

Knowledge

Speech Acts

Context


Intentions,Services

Intentions,Services

Reasoning

Reasoning

Reasoning

Reasoning


Shared Domain Knowledge


Collaboration, Role

REACTIVE

COGNITIVE

ENVIRONMENT




ENVIRONMENT

ENVIRONMENT

ENVIRONMENT


Agent Cognitive Semantic Model

Context(Situational Knowledge, Constraints)

Semantic KnowledgeIntentions

(Tasks, Workflows, Services)

Goals(Objectives, Guidance)

Intensional LogicalReasoning

(Decisions, Inferences)

Atomic Actions


Perceptions

EnvironmentData, Sensors

CommunicatingSpeechActs

CommunicatingSpeechActs


Implicit Semantic Knowledge

Goals

Knowledge

Speech Acts

Context

Goals

Knowledge

Speech Acts

Context

Purposeful Communications

Intention IntentionRequest, Committment

Shared Knowledge


Collaboration, Role

REACTIVE

COGNITIVE

Typically implemented via app specific

protocols, not universal

Usually defined by very few app specific msg types, not universal

Implicit semantic models by system designer, at most

explicit data element structure.

Never explicitly defined in system, only

implicitly by requirements

Never explicitly defined in system, only

implicitly by requirements


Explicit Services and Universal Speech Acts, No Explicit Semantics

Goals

Knowledge

Speech Acts

Context

Goals

Knowledge

Speech Acts

Context


Intentions,Services

Intentions,Services

Reasoning

Reasoning

Reasoning

Reasoning


Shared Knowledge


Collaboration, Role

REACTIVE

COGNITIVE

ENVIRONMENT




ENVIRONMENT

ENVIRONMENT

ENVIRONMENT


Explicit Semantic Knowledge, Services, and Speech Acts

Goals

Knowledge

Speech Acts

Context

Goals

Knowledge

Speech Acts

Context


Intentions,Services

Intentions,Services

Reasoning

Reasoning

Reasoning

Reasoning


Shared Knowledge


Collaboration, Role

REACTIVE

COGNITIVE

ENVIRONMENT




ENVIRONMENT

ENVIRONMENT

ENVIRONMENT


Explicit Context Knowledge

Goals

Knowledge

Speech Acts

Context

Goals

Knowledge

Speech Acts

Context


Intentions,Services

Intentions,Services

Reasoning

Reasoning

Reasoning

Reasoning


Shared Knowledge


Collaboration, Role

REACTIVE

COGNITIVE

ENVIRONMENT




ENVIRONMENT

ENVIRONMENT

ENVIRONMENT


Explicit Goal Knowledge

Goals

Knowledge

Speech Acts

Context

Goals

Knowledge

Speech Acts

Context


Intentions,Services

Intentions,Services

Reasoning

Reasoning

Reasoning

Reasoning


Shared Knowledge


Collaboration, Role

REACTIVE

COGNITIVE

ENVIRONMENT




ENVIRONMENT

ENVIRONMENT

ENVIRONMENT


NCOIC Integrated Ontology


NCOIC Integrated Knowledge Base – An NCOIC Ontology

• Create an integrated NCOIC knowledge Base that can be used by customers and member companies

• Create an NCOIC ontology that can be constructed from FT and WG ontologies to unify the NCOIC work products

– Provide a map of Network Centric Operation aspects– Incorporate NCOIC Lexicon– Capture descriptive knowledge about NCO aspects

• Map current NCOIC efforts against NCOIC ontology– To provide a context for research efforts and discussion– To identify shortcomings and candidate areas for research

• Enable evaluation of Customer Requirements and force initiatives against Net Centric Aspects and NCOIC work products

• Identify Specify Interoperability patterns, their structural solutions and their relationship to NCOIC work products

• Enable characterization of each solution using NCO evaluative and descriptive models

• Create manageable and scalable NCOIC ontology that can evolve• Expand to capture and influence Customer requirements specifications• Capture the operational space


NCOIC SII WG Work ProductKnowledge base

• Each WG Product has a document and associated Ontology to enable incorporation into a larger model

• SII WG Product Concepts– SII Integrated Ontology enable dependent relationships to be

made between:• NCO Tenets, Reference Models, • NCO SCOPE Model and its Descriptive Dimensions• Interoperability Causes• Interoperability Patterns – Focus on Service and Information • PFCs• Customer Requirements and Capabilities• Open Standards• Able to be component part of NCOIC Level Integrated Knowledge

Base and Ontology


Approach

NCOICIntegrated KB and Ontologies

Customers- ETE Capabilities

Architects- Patterns

-SCOPE Model- Interop Problems

Engineers- PFCs

- Profiles

Vendors- COTs/Gots

FT & WGs

Tools (Industry, Vendor, Govt.)

Govt and Member Companies, Prod Vendors


NCOIC Product Map (SII WG Perspective)

NCAT

SIIIntegrated

Knowledge BasePFCs

SII WGInteroperabilityPatterns-Information Exchange, Semantics-Services, Mediation-Msg Content Transformation-Collaboration, Workflow-Discovery, Context Awareness-Autonomicity, Management

Customer Reqts-DAR,CADM, DAP-DoDAF to DRL-JCIDS-PIM-NCOW RM-Capital Planning-PPBE-Acquisition-BCIDS-Net Ready KPPs-KIPs-DISRonlineProfiles

NCOICIntegratedKnowledge

Base

NCOICSCOPE

Document

NCOICLexicon

NCOICIntegratedOntology

NCOTenets

Ontology

Mobility

SIIIntegratedOntology

InteroperabilityCauses

Document

InteropOntology

SCOPEOntology

NCO TenetOntology

OpenStandards

OpenStandardsOntology

CustReqts

Ontology

PFCOntology

NCO InteropPattern

Ontology

IA


Integrated NCOIC Product Map (Proposal)

NCAT

PFCsInteroperabilityPatterns-Information Exchange, Semantics-Services, Mediation-Msg Content Transformation-Collaboration, Workflow-Discovery, Context Awareness-Autonomicity, Management

Customer Reqts-DAR,CADM, DAP-DoDAF to DRL-JCIDS-PIM-NCOW RM-Capital Planning-PPBE-Acquisition-BCIDS-Net Ready KPPs-KIPs-DISRonlineProfiles

NCOICIntegratedKnowledge

Base

NCOICSCOPE

Document

NCOICLexicon

NCOICIntegratedOntology

NCOTenets

Ontology

Mobility

InteroperabilityCauses

Document

InteropOntology

SCOPEOntology

NCO TenetOntology

OpenStandards

OpenStandardsOntology

CustReqts

Ontology

PFCOntologyIA

NCO InteropPattern

Ontology

BuildingBlocks

BuildingBlocks

Ontology

Interactions between WGsAnd FTs not defined here


Recommendations• Unify the NCOIC knowledge and products using the

NCOIC KB and Ontologies – Align with similar customer efforts in KB– Foundation for collaborative engineering

• Establish a group to manage the NCOIC Ontology and KB– Each WG has one focal person for input and vetting– currently being done by SII WG

• Training for Semantic Information Capture• Tools and commercial hosting platforms for NCOIC

ontology (Infrastructure Recommendation)– Assist NCOIC marketing efforts– Assist engineering efforts– Budget (plan to follow)


Emergency Disaster Response Information Coordination

Semantics


“Many of the problems we have identified can be categorized as “information gaps” – or at least problems with information-related implications, or failures to actdecisively because information was sketchy at best. Better information would have been an optimal weapon against Katrina. Information sent to the right people at the right place at the right time. Information moved within agencies, across departments, and between jurisdictions of government as well. Seamlessly. Securely. Efficiently.”

The Final Report of the Select Bipartisan Committee to Investigate the Preparation for and Response to

Hurricane Katrina


Coordination Problems

• Lack of organized information focusing on coordination activities and status:– Resources– Participants– Incident resolution

• No Common Operating Picture relating evolving overall coordination situation.

• Inability to plan specific coordination activities for different disaster scenarios


Information and Communication Problems

• Focus on data elements rather than model structure and domain

• Messages only related to each other by message ID; i.e. patterns of coordination not readily apparent

• Semantic descriptions of data elements in message schemas inadequate

• No standards used to represent higher levels of semantic expressiveness in data model, e.g. RDF, OWL

• Architecture does not specify how information sharing takes place among responders in any dynamic or adaptive manner

• No directory structure exists within DMIS to enable service discovery


Project• Research and Development of Emergency Disaster Response

Information Coordination Semantic (ED-RICS) framework to improve emergency response coordination

• Focus on semantic architectural model that creates common operating picture (COP) of evolving emergency response coordination situation

• Represent COP as set of discrete semantic coordination patterns (SCP) derived from XML emergency messages

• Ontology based network coordination situation analysis identifying coordination anomalies, completion states, resource commitments, and incident focus problems

• Technologies include:– EDXL and CAP alert message standards– DHS NRP scenarios– Protégé 2000 ontology tool with OWL plugin– Domain ontologies with non-programmatic concept inferences– Web services– Concepts from knowledge representation and descriptive logic


DHS-NRP Scenario Analysis• Scenario 1: Nuclear Detonation – 10-Kiloton Improvised Nuclear Device• Scenario 2: Biological Attack – Aerosol Anthrax• Scenario 3: Biological Disease Outbreak – Pandemic Influenza• Scenario 4: Biological Attack – Plague• Scenario 5: Chemical Attack – Blister Agent• Scenario 6: Chemical Attack – Toxic Industrial Chemicals• Scenario 7: Chemical Attack – Nerve Agent• Scenario 8: Chemical Attack – Chlorine Tank Explosion • Scenario 10: Natural Disaster – Major Hurricane• Scenario 11: Radiological Attack – Radiological Dispersal Devices• Scenario 12: Explosives Attack – Bombing Using Improvised Explosive

Devices• Scenario 13: Biological Attack – Food Contamination• Scenario 14: Biological Attack – Foreign Animal Disease (Foot and Mouth

Disease)


Current Emergency Disaster Response Information Interoperability Network

Common Alerting Protocol - CAP

Emergency Data Exchange Language - EDXL

SOAP, WSDL, HTTP

National Information Exchange Model (NIEM)

Disaster Management Interoperability Service (DMIS)

Emergency Provider Access Directory (EPAD)Services

Data Model

Messages

Disaster Management Interoperability Services (DMIS)

Emergency Provider Access Directory (EPAD)


Emergency Messaging Languages

• CAP: emergency messaging standard used to alert responders and public in general of emergency situations as they occur.

• EDXL-RM: messaging standard used to convey information regarding emergency specific resources.

• EDXL-DE: emergency messaging standard used as container for CAP and EDXL-RM messages. EDXL-DE may also contain emergency data not otherwise included in CAP or EDXL-RM messages.


ED-RICS Capabilities

Provides universal shared information analysis through creation of common operating picture (COP) of all coordination activities, including:

• Committed resources• Responder locations with respect to incident area • Coordination activity completion status• Anomaly analysis, such as overcommitted resources, etc.• Interactive execution environment between knowledge

framework and responders, response managers, messaging systems, databases, and other personnel and systems


Coordination Intersection

Information

Resources

Activities

Messages

ED-RICS


Emergency Disaster Response Services and Information Framework

Common Alerting Protocol - CAP

Emergency Data Exchange Language - EDXL

NIEM

SOAP, WSDL, HTTP

(DMIS)

(EPADS)

Emergency Situational Information

Service•Plan Management•Status Monitoring•Situation Analysis

•Anomalies Identification

Web Services

Data Models

Current

New

EDXL-RM OWL

EDXL-DE OWL

CAP OWL

Semantic

Semantic Data Model for Emergency Disaster

Planning, Monitoring, Analysis

semantic interoperability1 semantic interoperability net centric perspective presented to sicop team...

Documents

semantic models

semantic interoperability2

semantic breakdown slide

semantic policy framework

semantic service descriptions

information gaps

better information

self management