Von Schweber Living Systems

1

Dynamically Synthesizing Systems of Systems

© Synsyta 2006. All Rights Reserved.

Synsyta and Neological Proprietary

Living Systems &Living, Liaising Languages

Key to Netcentric Interoperability

January 24, 2006

Erick Von SchweberCTO and Chief [email protected]

2© Synsyta 2006. All Rights Reserved.


Riding the Waves

First WaveUniprocessing

Second WaveParallel & Distributed

Processing

IBMOn-Demand

Sun N1

HP AdaptiveEnterprise

BladeServersSMTA &

Multicore

WebGrids

SemanticWeb

SOA

P2P

Web 2.0

AJAX

RSS

MDA

Globus

Model DrivenSemantic GridWireless

MeshAd Hoc

Networks

WebServices

Third WaveComputing

Fabrics

RDFOWL

Living, Organic Architecture Systems

into EcosystemsCultural Co-Evolution

SymbiosisSynsytia Surveying

Living Liaising Languages

Semantic Index GridManaged Logic

ccNUMA clusters

Reconfigurable Logic

FPGAs

XML

The challenge of our eraTransforming clockwork mechanism into living

technology



Outline

• Riding the waves• Biologically and culturally inspired

interoperation• Foundations• Roadmap• Next steps



A Human+Machine culture ofLiving, Liaising Languages



Machine-Machine vs. Human–Human Communication

Machine-Machine Human–HumanThey are programmed in advance of interoperation

We learn each other’s lingos as we communicate

They require instructions We point to examples

Ambiguity simply does not compute

We use analogy, metaphor and conceptual blending

They stonewall We negotiate

They are static and rigid We self-adapt

Their languages change with periodic releases and revisions

Our languages are in constant flux

They must be reengineered We culturally evolve



RoadmapSecond Stage

Collaborative ontologies

andComposable languages

Third Stage

Living, liaising language hubs

First Stage

Automated interoperation of heterogeneous languages and

ontologies

Fourth Stage

Culturalco-evolution

& symbiosis of living, liaising

languages

Composable theories extend collaborative classifications

Increasing collaboration and generality



Foundations

• NeoLogical SURVEYOR, 1983 – present• Computing Fabrics, 1996 - present• “Software Services Grid” workshop, July 2001• “Model Driven Semantic Grid” GGF2, July 2001• “Information Representation & Transformation” for DARPA TIA

– “Model Driven Semantic Grid, and Beyond” workshop, Jan. 2003– “How the Stacks Stack Up” workshop, June 2003– “Integrated Semantics” workshop, August 2003

• “Managed Logic worked example” white paper 2004• “Roadmap for Semantics in netcentric Enterprise Architecture” white paper

for GSA, August 2005 - present

• Ken Baclawski• Desmond D’Souza• Dave Frankel• Elisa Kendal• Robert Kent

• Deb McGuinness• Sheila McIlraith• Jeff Smith• John Sowa

Cast of Characters(frequently appearing)

• Erick Von Schweber• Linda Von Schweber• Cory Casanave• Joseph Goguen



Surveying via unknown dimensions and attributes

supported by the ability to learn, decide and adapt

A knowledge horizon is the boundary of what we can knowgiven what we already know and the capabilities we possess.

It is as far as our knowledge and tools will permit us to see.

Searchingvia known

dimensions and attributes

Browsingvia links

Retrieving via unique identifiers

(a link or key)

Knowledge already in

your head or system



SURVEYOR zeros in on the optimal knowledge context & focuseven when this lies beyond the knowledge horizon

Context Focus



Information Flow & InteroperabilityStudy on Intercommunity Intelligence Information Flow (I3F)DARPA IAO 2003

A community of service components afloat in a sea of representations

IDEFx PSLExpress

BPELBPML

FOL HOL

Modal TemporalProbabilisticNon-Monotonic

ZedSlang

Category Theory

KIF

CGsCycL

CL RuleML

OWL

RDF(S)DAML+OIL

Topic Maps

XML

NETLFRLOKBC UML

SQL

MOF

CWM

MDSOC

ORMCMESIAM

EELD



Objective:

Deliver the expressivity and reasoning power of Logic based Languages in a modular fashion with MDA’s modeling, transformation, and management support and the Semantic Web’s markup

– Represent each formalism as a componentized metamodel

– Define an open formal structure of metamodels & reusable metamodel components: structural, logical & semantic

– Support composition & weaving of metamodel components

– Exploit MDA machinery to conduct transformations between the metamodels and to conduct mappings to and from concrete syntaxes (syntax agnosticism)

– Formally but agnostically ground MDA

– Integrate modeling tools, repositories, transformation machinery, generators, inference engines, …, and the communities

The MDA Stack

MOF2(MDA Common Core)

UML

UML Models

UML Instances

CWM

CWMModels

CWMInstances

XML, XML Schema, &

XMLns

RDF(S)(Notation 3)

OWL(~DAML+OIL)

Rich Logics & Reasoning

The Semantic Web Stack

OWL

FOL

Common Logic

KIF

C G s

HOL

CycL

CL-X

CGIF

CLML

The Logic Based Language Stack

The Information Flow Framework

Top Metalevel

Upper Metalevel

Lower Metalevel

Category Theory(meta) Ontology

Upper Classification(meta) Ontology

Upper Core(meta) Ontology

Top Core(meta) Ontology

Lower Core(meta) Ontology

Lower Classification(meta) Ontology

Ontology(meta) Ontology

Model Theory(meta) Ontology

Algebraic Theory(meta) Ontology

How the Stacks Stack Up June 2003Study on Intercommunity Intelligence Information Flow (I3F) DARPA IAO

StacksTasks

MDA Semantic Web

CL IFF

Modeling, Transformation & Management Machinery

MOF XMI UML CWM

Tractable Expressivity & Markup

OWL-DLRDF(S) XML

Maximal Expressivity, Reasoning & Formality

KIF -> CL

Formal Grounding, Interoperability & Composition

CT IFFFCA

No single formalism will do, so how do we Integrate the Stacks?



Roadmap – First StageSecond Stage



Third Stage


First Stage


ontologies

• OWL-RA as meta language Relative metamodeling Architecture RA language idioms Powertypes Term-Concept map• Semantic Core language hub• IF classifications & infomorphisms• Chu Spaces & Transforms• Galois Lattices & Formal Concept Analysis• Relational DBMS

• Progressive & Collaborative mapping using SURVEYOR Think of “formal Flicker”• Formal instances for hub languages & ontologies• Language concern dimensions ontology Like IFF LoT• ε-Connections between language ontology components

• Chu Spaces/Transforms lifted to Institutions & Institution Morphisms in two phases• Institutionalized language concern dimension ontology

Fourth Stage



languages

• Lifting of Institutions to Charters & Parchments• Model theoretic semantics as a composable concern dimension• Open-ended collection of multiple meta-mathematics





The Example ProblemUML OWL-DL

OBI UBL<<ontologyClass>>

RosettaNet OBIOrder Item.:class

<<ontologyClass>>PO101 Assigned Identification:slotRelation

<<ontologyClass>>PO107 Product/ServiceID:slotRelation

<<ontologyClass>>PO104 Unit Price:slotRelation

<<ontologyClass>>PO102 Quantity Ordered:slotRelation

<<ontologyClass>>RosettaNet OBIOrder Item.:class










<<ontologyClass>>UBL Order Line Item.:class

<<ontologyClass>>BuyersID:slotRelation

<<ontologyClass>>ID:slotRelation

<<ontologyClass>>Description:slotRelation

<<ontologyClass>>Quantity:slotRelation

<<ontologyClass>>PriceAmount:slotRelation

<<ontologyClass>>LineExtensionAmount:function















RosettaNet_OBI_Item_ODMxUBL_Item_ODM:TransformationMap

ItemMap:ClassifierMap

ItemDetailsMap:ClassifierMap

OwnedElemnt

OwnedElement

BuyersID = PO101Assigned Identification;ID = PO107 Product/ServiceID;

Quantity = PO102 Quantity Ordered;PriceAmount = PO104 Unit Price:

ProcedureExpression

functionsource

source

source

source

source target

target

target

target

target

target

target





















Heterogeneous• Schemas• Models• Ontologies• Schema languages• Modeling languages• Ontology languages



Automated Interoperation of Heterogeneous Languages and Ontologies

Meta Language• OWL-RA as meta language

– Relative metamodeling Architecture• Language idioms

– Powertypes– Term-Concept map

Language hub

• Semantic Core

Mathematics• Information Flow (IF)

– Classifications and infomorphisms• Chu Spaces and Chu Transforms• Galois Lattices and Formal Concept

Analysis

RDBMS, SQL &NeoLogical SURVEYOR



Isomorphic Mathematical Approaches

Types

Tokens

=

Types

Tokens

=

Classification A Classification B

Infomorphism:a pair of adjoint

functionsInformation Flow

(IF)

States

Objects

Chu Space A

States

Objects

Chu Space B

Chu TransformChu Spaces



Language MapChu Spaces for UML Class and OWL-DL

Source Chu Space UML_Class UML_Attribute U_is_related_to

PO 1 0 0POLineItem 1 0 0POLineItem_Attribute 0 1 0POLineItem is_related_to

POLineItem_Attribute0 0 1

Target Chu Space OWL-DL_Class OWL-DL_Property O_is_related_to

PO 1 0 0POLineItem 1 0 0POLineItem_Attribute 1 0 0POLineItem is_related_to

POLineItem_Attribute0 1 1



Language MapMerged Chu Space: UML Class and OWL-DL

Merged Chu Space: UML Class and OWL-DL

UML_Class

UML_Attribute

U_is_related_to

OWL-DL_Class

OWL-DLProperty

O_is_related_to

PO 1 0 0 1 0 0POLineItem 1 0 0 1 0 0POLineItem_Attribut

e0 1 0 1 0 0

POLineItem is related to POLineItem_Attribute

0 0 1 0 1 1

Merge duplicate columnsMerge duplicate rows

Triangulate

Reduced Chu Space OWL-DL_Class

UML_Class

UML_Attribute

U_is_related_to

OWL-DL_Property

O_is_related_toPOPOLineItem

1 1 0 0

POLineItem_Attribute 1 0 1 0

POLineItemis related to POLineItem_Attribute

0 0 0 1


Synsyta and Neological ProprietaryLanguage MapHasse diagram of the Galois latticeover the merged Chu Space

I

OWL-DL_ClassOWL-DL_PropertyU_is_related_toO_is_related_to

UML_Class UML_Attribute

Φ

POLineItem is_related_to POLineItem_Attribute

POLineItem_Attribute

POPOLineItem

POLineItem_Attribute

POPOLineItem

Also called aConcept Lattice by

Formal Concept Analysis (FCA)



Language MapLanguage Map of the Chu Transform

Language MapCLTup : UML_Class OWL-DL_ClassCLTup : UML_Attribute OWL-DL_ClassCLTup : U_is_related_to(UML_Class, UML_Attribute) O_related_to(OWL-DL_Class,

OWL-DL_Class) {where type(O_related_to) = OWL-DL_Property

The language map is one of the pair of contravariant functions that constitute

an infomorphism inInformation Flow (IF)



Language MapChu Transform as a Chu Space

Generalized source Chu Space for UML class

UML_Class

UML_Attribute

owns

S1, i.e., a UML class

1 0 0

S2, i.e., a UML attribute

0 1 0

S3, i.e., the owning of a UML attribute by a UML class

0 0 1

Generalized Chu Transform UML to OWL-DL(as Chu Space)

OWL-DL_Class

OWL-DL_Property

S1, i.e., a UML class 1 0

S2, i.e., a UML attribute

1 0

S3, i.e., the owning of a UML attribute by a UML class

0 1

Generalized targetChu Space for OWL-

DL class & property

OWL-DL_Class

OWL-DL_Property

T1, i.e., an OWL-DL ontology class

1 0

T2, i.e., an OWL-DL property

0 1


Synsyta and Neological ProprietaryDomain MapUML class model of RosettaNet Open Buying onInternet (OBI) Order Item and instance

OBI Order Item

PO101 Assigned IdentifierPO107 Product/ServiceIDPO104 Unit PricePO102 Quantity Ordered

PO1138_1 : OBI Order Item

PO101 Assigned Identifier = Cust_123PO107 Product/ServiceID = SKU_abcPO104 Unit Price = $10PO102 Quantity Ordered = 5 units


Synsyta and Neological ProprietaryDomain MapOWL-DL ontology of Universal Business Language(UBL) Order Line Item and instances

UBL Order Line Item BuyersID

ID

Description

Quantity

PriceAmount

LineExtensionAmount

has_element

<rdf:Description rdf:about="PO1138_1"><rdf:type rdf:resource="UBL_Order_Line_Item"/></rdf:Description><rdf:Description rdf:about="Cust_123"><rdf:type rdf:resource="BuyersID"/></rdf:Description><rdf:Description rdf:about="SKU_abc"><rdf:type rdf:resource="ID"/></rdf:Description><rdf:Description rdf:about="$10"><rdf:type rdf:resource="Quantity"/></rdf:Description><rdf:Description rdf:about="5 units"><rdf:type rdf:resource="PriceAmount"/></rdf:Description><UBLOrder_Line_Item rdf:about="PO1138_1> <has_element rdf:about="Cust_123"/> <has_element rdf:about="SKU_abc"/> <has_element rdf:about="$10"/> <has_element rdf:about="5 units"/></UBLOrder_Line_Item>



Domain map in the context of the Language map

• The language map creates two partitions for the Domain map– The class-attribute/class partition– The owns / property partition

• The Domain map consequently must have two component maps– A component map for Classes and Attributes– A component map for relations



Domain MapChu Spaces for OBI and UBL Classes & Attributes

Chu Space of OBIClasses & Attributes

OBIOrderItem

PO10AssignedIdentifier

PO107Product/ServiceID

PO104 UnitPrice

PO102QuantityOrdered

PO1138_1 1 0 0 0 0Cust_123 0 1 0 0 0SKU_abc 0 0 1 0 0$10 0 0 0 1 05 units 0 0 0 0 1

Chu Space of UBL

Classes

UBLOrderLineItem

BuyersID ID Description PriceAmount

Quantity LineExtensionAmount

PO1138_1 1 0 0 0 0 0 0Cust_123 0 1 0 0 0 0 0

SKU_abc 0 0 1 0 0 0 0

$10 0 0 0 0 1 0 0

5 units 0 0 0 0 0 1 0



Domain MapChu Spaces for OBI and UBL relations

Chu Space of OBI UML class – attribute relations

owns

(PO1138_1, Cust_123) 1

(PO1138_1, SKU_abc) 1

(PO1138_1, $10) 1

(PO1138_1, 5 units) 1

Chu Space of UBLOWL-DL property instances

has_element

(PO1138_1, Cust_123) 1

(PO1138_1, SKU_abc) 1

(PO1138_1, $10) 1

(PO1138_1, 5 units) 1


Synsyta and Neological ProprietaryDomain MapMerged Chu Spaces for OBI and UBLclasses & attributes

MergedClass-attr/ClassChuSpace

OBIOrderItem

PO101AssignedIdentifier

PO107Product/ServiceID

PO104UnitPrice

PO102QuantityOrdered

UBLOrderLineItem

BuyersID ID PriceAmount

Quantity

PO1138_1 1 0 0 0 0 1 0 0 0 0

Cust_123 0 1 0 0 0 0 1 0 0 0SKU_abc 0 0 1 0 0 0 0 1 0 0$10 0 0 0 1 0 0 0 0 1 05 units 0 0 0 0 1 0 0 0 0 1



Domain MapChu Spaces for OBI and UBL relations

Merged owns/property Chu Space owns has_element

(PO1138_1, Cust_123) 1 1(PO1138_1, SKU_abc) 1 1(PO1138_1, $10) 1 1(PO1138_1, 5 units) 1 1



Domain MapHasse diagram of the Class-Attribute Lattice

Φ

I

OBI Order Item

UBL Order Line

Item

PO104 Unit Price

Price Amount

PO107 Product/ServiceID

ID

PO101 AssignedIdentifier

BuyersId

PO102 QuantityOrdered

Quantity



Domain MapHasse diagram of the Relation Lattice

Φ

ownshas_element

I



Language and Domain Maps

Domain MapCDTup : OBI Order Item UBL Order Line ItemCDTup : PO101Assigned Identifier BuyersID CDTup : PO107 Product/ServiceID ID CDTup : PO102 Quantity Ordered Quantity CDTup : PO104 Unit Price PriceAmount CDTup : owns has_element

Language MapCLTup : UML_Class OWL-DL_ClassCLTup : UML_Attribute OWL-DL_ClassCLTup : U_is_related_to(UML_Class, UML_Attribute)

O_related_to(OWL-DL_Class, OWL-DL_Class) {where type(O_related_to) = OWL-DL_Property

CLT:= Chu Language TransformCDT:= Chu Domain Transform (up:= type to type map)



Extending the Example Mapping Scenario

Language 1 Language 2

Ontology 1a Ontology 2a

Model 1ai Model 2ai

Data 1ai Data 2ai



Hub & Spoke Scenarios also Supported

SpokeLanguage

HubLanguage

SpokeLanguage

SpokeOntology

HubOntology

SpokeOntologySpoke

OntologySpokeOntology

HubOntologyHub

OntologySpoke

OntologySpokeOntology

SpokeModel

SpokeModelSpoke

ModelSpokeModelSpoke




ModelSpokeModel



Challenges of the Second Stage

• BIG ontologies• Varying ontologies• Collaborative ontology development• “Our” ontology as a dynamic merging of “our” ontologies • Ontologies that come and go without notice• Acquiring common ontology instances across ontologies

(motivated by the methodology of the First Stage)

Ontology mapping as a routine, integral,

every day practice



Roadmap - Second StageSecond Stage



Third Stage


First Stage


ontologies


• Progressive & Collaborative mapping using SURVEYOR Think of “formal Flicker”• Formal instances for mapping hub languages & ontologies• Ontology of language concern dimensions (like IFF LoT at the language definition level)• ε-Connections between language ontology components


Fourth Stage



languages






Second Stage

Collaborative ontologiesand

Composable languages

• Progressive & Collaborative mapping using SURVEYORTM

Think of “formal Flicker”• Formal instances for mapping hub languages & ontologies• Ontology of language concern dimensions (like IFF LoT at the language definition level)• ε-Connections between language ontology components



Overlapping knowledgeCommon knowledge horizon

Overlapping knowledgeOverlapping knowledge

horizons

Disjoint knowledgeOverlapping knowledge horizons

Other is beyond knowledge horizon

Disjoint knowledgeDisjoint knowledge horizons

Disjoint knowledgeOverlapping knowledge

horizonsOther is partly within knowledge horizon

Concentric knowledge

Disjoint knowledgeOverlapping knowledge

horizons

Concentric knowledgeOther is partly beyond

knowledge horizon

Knowledge Horizons in a systems context

Overlapping knowledgeOverlapping knowledge

horizonsOther is partly beyond

knowledge horizon

CurrentKnowledg

e

Knowledge

Horizon

Surveyor. © 2002-6 by Infomaniacs/Neological. All Rights Reserved. 36



Roadmap - Third StageSecond Stage



Third Stage


First Stage


ontologies




Fourth Stage



languages






MAGIC - Managed Logic

Automates heterogeneous interoperability between:• New Activities by defining end-points with MAGIC • Existing Activities by redefining end-points with MAGICTasks/communities may be uncoupled or loosely coupled

Producer defines task & community specificlanguagewith MAGIC

MAGIC

Consumer defines distinct task &

community specificlanguage

with MAGIC

MAGIC

Information flows overbridge generated andmanaged by MAGIC

MAGIC MAGIC

MAGIC is not yet another language - it is a language machine



Managed LogicIn

form

atio

n Fl

ow

MAGIC - Managed LogicLifecycle Management of Synthetic Language Systems and their Interrelations

• XML• RDF(S)• Topic Maps• DAML+OIL & OWL• OKBC• CycL• KIF• Common Logic & SCL• Conceptual Graphs• UML• DSLs• IDEFx & Express• SQL• BPML and BPEL• VHDL & Verilog• ADLs

Example SLS

Core Capabilities

• Formally Define SLS– Componentize existing SLS– Compose new general

purpose and domain specific SLS

– Modify & hybridize SLS to meet new and changing requirements

• Coordinate Multiple SLS– Synergistically employ

multiple, diverse SLS (transoperate)

– Interoperate between distinct or versioned SLSLanguage

Components

VerificationFormal

LanguageDefinitions

Syntactic / SemanticLanguage

Transformations

Formal Composition

Automated Production

Composition

Query/Browse/Trace

Define/Modify/Verify

Transformation

Federation

Import/Export

Code Generation

Code Deployment

Code Sustainment

ServicesDecomposition

Version Mgt.

Services encapsulate formal underpinnings

Informal& Formal

SLSSpecs

Formal Aspect

Oriented Decompositio

n• Modeling• Specification• Knowledge Representation• Learning

• Engineering• Ontology• Reasoning• Integration & Interchange

SLS are used for …

The facility does not require an intermediate canonical

language;the library spans

complementary language components in the spirit of

Sowa’s Knowledge Soup and Kent’s Information Flow

Framework.

Enables Flow

SLS (Synthetic Language Systems) are logics, representational formalisms, formal languages and domain specific languages taken together with their

associated products, e.g., expressions, instances, models and ontologies



Institution Example

Collie DogFido Dog

Fido Collie

FidoDog

Collie

Functor injects logical symbols and guarantees well-formedness of sentences, i.e., grammar

Sets of ModelsSentences

Signatures (syntax)

DogCollie[Fido]

This example applies the theory of Institutions at an ontological level; Institutions are customarily used to abstract logics and languages

∩∩

∩

=

States

Objects

States

ObjectsChu Transform

Chu Space A Chu Space BTypes

Tokens

==

Types

Tokens

==

Classification B

Infomorphism:a pair of adjoint

functions

Classification A IF ChuFor comparison



Roadmap – Fourth StageSecond Stage



Third Stage


First Stage


ontologies




Fourth Stage



languages






Next steps

• First Stage - Proof of Concept– Run real mapping examples in RDBMS using SURVEYOR and

advanced maths• Second Stage - R&D

– Collaborative, progressive ontology mapping using SURVEYOR and Web 2.0 technologies• Communities of ontologies; emergent “consensus” semantics

– Ontology of language concern dimensions as meta ontology to Semantic Core• Composable languages



SURVEYORDemo

Education

Von Schweber Living Systems