The Special Role of M&S in Cross-COI Mediation

Bernard P. Zeigler, Ph.D.,Arizona Center for Integrative Modeling and Simulation

andJoint Interoperability Test Command

Fort Huachuca, AZ 85613-7051zeigler@ece.arizona.edu

System Entity Structure

• The M&S COI has a special role to play in the area of mediation among metadata schemes developed by the various COIs

• Interoperability among the web services of COIs will require a more dynamic elements and associated simulation interpretations, typical of the M&S concerns

• Use System Entity Structure (SES) from Theory of Modeling and Simulation

• The SES is the basis for a methodology, and practical tool set that supports data engineering and modeling of application domains with hierarchical system characteristics

• It is being applied by the Joint Interoperability Test Command (JITC) to imagery metadata characterization for the some of the sensor products of the National Geospatial-Intelligence Agency (NGA).

SES Overview

• SES methodology and tool set– Java classes using Sun’s XML classes– develop well-structured, hierarchically composable data models– can be mapped to a variety of legacy formats.

• Due to its fairly simple axiomatic structure the SES lends itself to tool-supported structure analysis– tools to compute the common elements and substructures of a

pair of SES representations. – not as powerful as a full-blown, logic-based ontology framework– it is sufficiently expressive for real application – sufficiently well structured to allow useful tool support.

System Entity Structure –Basic Concepts

• Entity – a thing in the real world, e.g book• Variable – an attribute of an entity, e.g. title• Aspect – decomposition, a way to break down an entity into parts or

components (entities), e.g., front cover, back cover, pages• Specialization – a classification, a way to classify an entity into special

cases or subclasses, e.g., genre• MultipleAspect – a decomposition into similar parts, e.g. pages

• Entities alternate with Aspects and Specializations• Prunings generate family of pruned entity structures

(PES)

Example: SES Representation of Book

book

contentDec

chapters

chapter

chapterDec

explanation

theorem

proof

theorems

example

physicalDec

pages

page

front cover

back cover

Aspect

Entity

MultiAspect

genre

bio-graphy

fiction

Specialization

physicalDec

pages

page1

front cover

back cover

fiction_book

page100

pruning

Generation of XML Schema

<?xml version="1.0" encoding="UTF-8"?><!DOCTYPE entity SYSTEM "ses.dtd" []><entity name="House">

<aspect name="roomDec"><entity name="Kitchen"></entity><entity name="LivingRoom"></entity><entity name="DiningRoom"></entity>

</aspect><var name="address" rangeSpec="string"></var><var name="registrationNumber" rangeSpec="decimal

number"></var></entity>

<?xml version="1.0" encoding="UTF-8"?><!DOCTYPE House SYSTEM "HouseDTD.dtd" []><House ="6131 E. San Leandro St" registrationNumber="10678">

<aspectsOfHouse><roomDec>

<Kitchen></Kitchen><LivingRoom></LivingRoom><DiningRoom></DiningRoom>

</roomDec></aspectsOfHouse>

</House>

SES for House based on ses.dtd

(sesForHouse.xml)HouseDTD.dtd

PES for House based on HouseDTD.dtd

(SanLeandroHouse.xml)

merge SES

merged SES

merge PESmerged

PES

Synthesizing Large SESs and PESs From Components

Top Level SES Representation of NITF Core Information

Mission

Mission Image Aspect

Image Aspect

Images

Image

Identification Segment descriptions Ground referenceSensor

Image Data Aspect

SES Breakdown of Identification Information

Identification

Identification Data Aspect

Image Sequence Data Aspect

Sequence ID~ISID

Original Image Data Image Sequence

~NRG~NCG~TRG~TCG

Date/Time Data

Simultaneous Collection

Mission

Mission Image Aspect

Image Aspect

Images

Image

Identification Segment descriptions Ground referenceSensor

Image Data Aspect

Identification Entity Merged Into Mission SES

Identification

Identification Data Aspect

Image Sequence Data Aspect

Sequence ID~ISID

Original Image Data Image Sequence

~NRG~NCG~TRG~TCG

Date/Time Data

Simultaneous Collection

Advantages of Hierarchical Methodology for Managing Body of Standards

• Divide and Conquer - Complex standard is decomposed into smaller more manageable pieces

• Pieces can be developed and maintained individually as modifications are required

• Builds (merges) can be performed at any time to create current version of overall standard

• Some pieces can be stand-alone standards under the same or other authorities

• XML instances of component standards can be merged into larger XML instances of the overall standard.

Approach to Harmonization

• Multiple legacy metadata formats that partially overlap in their referents

• Attempt to find as much common core as possible

• Express common core in SES

• Expland SES with Specializations to capture non-core elements

• Re-derive original content of metadata by pruning from master SES

Tools for Harmonization

• Relational representation of SES• Commonality measures• Operations to increase commonality• Thesaurus

Relational Representation of SES

entities

SES can be represented as a collection of relations such as entityHasAspect, aspectHasEntity, entityHasSpecialization, etc. as shown. SES’s can be compared on the basis of their relational representation.

SES

aspects

multiaspects

specializations

attributes

entity

specialization

attribute multi-aspect0..n

10..n

aspect0..n

1 1..n

0..n

0..n

1..n

0..n

1 0..11..n

1..n

Finding common entities and SES’s

entities entities

commonentities

SES A SES B

Computing Commonality of SES’s

equality of subSES below commonentity ?

SES A

SES BsubSES below commonentity

common entity common

entity

Commonality = •fraction of shared entites•fraction of shared sub-SES’s

Entity Removal May Increase Commonality

House

couch

removeEntity (“LivingRoom”)House

sofa

House

House

LivingRoom

LivingRoom

Name Substitution may Increase Commonality

House

couch

addRename(“sofa”,”couch”)replaceAll()

House

sofa

House

couch

House

couch

Maintaining Thesaurus of Equivalent Names

thesaurusthesaurus.xml

store inpersistent form

addRename(“sofa”,”couch”)replaceAll()

House

sofa

House

couch

transferpairs

applyto SES

set representatives for equivalence classes

Representatives of equivalence classes are canonical names that will be used as defaults in application to SES’s, thereby providing a standard version of an SES.

Representative Class members

sofa couch, sofa, divan,..

wall side, wall,…

… …

Dynamic SES

• Need metadata that tracks processing as data is processed through stages

• Develop dynamic SES that can evolve through transformations

• Include forward and reverse links for traceability

Dynamic Process Example: Evolution of a Book

Book(concept)

Book(content

description) Book(physical

embodiment)

Book(re-printings)

Book(revisions)

Ex: Representing Versions In SES

book

contentDec

chapters

chapter

chapterDec

explanatory

theorem

proof example

theorems

engineering version –remove all proofs from theorems

mathematician’s version –remove all examples fromtheorems

theorem

example

theorems

theorem

proof

theorems

chapterDec

chapterDec

Trajectories in Space of SESs

BookSES

BookSES’

transformation

retain description of transformationto enable reverse traceability

Space of SES’sedges labeled by transformations and ancestor links

nodes are SES’s derived from precursors

Summary

• SES is rich enough to support approaches to harmonization and dynamic processing that will be involved in cross-COI mediation

• Representation as relations allow Java operation and measurement tools

• Supports alternative Mappings to XML to satisfy different applications

Bernard P. Zeiglerzeigler@ece.arizona.edu

ACIMSwww.acims.arizona.edu

JITCjitc.fhu.disa.mil

Contact:

More information:

Joint Interoperability, Assured Security, Best Value, Customer Success

Global Net-Centric Solutions -- The Warfighter's EdgeGlobal Net-Centric Solutions -- The Warfighter's Edge