Semantic Web Applications for Modeling and Simulation Lee W. Lacy Dynamics Research Corporation Captain Joel Pawloski U.S. Army TRAC-M July 11, 2001 DMSO Technical Exchange Meeting
Agenda Web Technology Evolution XML M&S Applications U.S. Army TRAC-M XML Research Semantic Web Background DARPA Semantic Web Research Potential Semantic Web M&S and C4I Applications
The Evolving Web 2010 2000 1990 Web of Knowledge Proof, Logic and Ontology Languages (e.g., DAML+OIL) Shared terms/terminology Machine-Machine communication 2010 Resource Description Framework (RDF) eXtensible Markup Language (XML) Self-Describing Documents 2000 HyperText Markup Language (HTML) HyperText Transfer Protocol (HTTP) Formatted Documents Foundation of the Current Web 1990 Based on Berners-Lee, Hendler; Nature, 2001
Hypertext Markup Language (HTML) Enabled standard communication of content combined with format Loosely defined specific (limited) grammar specified using SGML (specific language) Started by Tim Berners-Lee Standardized by World-Wide Web Consortium (W3C) Ubiquitous method of presenting and communicating data Used throughout DoD on both public internet and SIPRNET and other classified networks Not useful for machine search
eXtensible Markup Language (XML) Separates content from format (standard syntax) Simplified version of SGML (metalanguage) for defining eXtensible tag sets Started by Tim Bray et al based on conversations with Tim Berners-Lee Standardized by W3C Used to define updated HTML grammar (XHTML) Used in Modeling and Simulation community primarily for Data Interchange Formats (DIFs) that support data sharing
XML DIF Method ModSAF XML-based Data Interchange File (DIF) Janus CCTT
M&S and C4I use of XML HLA Data Interchange Formats (DIFs) JSIMS Common Component Workstation Battlespace Schema Combat XXI Scenario Files OneSAF Objective System Land Warrior User Interface Specifications Joint C4ISR Integration Facility (JCIF) Joint Battlespace Infosphere (JBI) XML Assessment
Specific U.S. Army M&S XML Research Initiatives Military Scenario Definition Language (MSDL) Computer Generated Forces (CGF) Behaviors Equipment Characteristics and Performance (C&P)
MSDL Objectives Support Scenario Developers Improve Scenario Quality Reduce Time to Develop Scenarios Reduce Costs of Developing Scenarios Support Scenario Generation Tool Developers Reduce Tool Development Costs Improve Tool Interoperability
Scenario Content Analysis Metadata Environment Organization Simulation Object Information Item Events Unassigned
MSDL Standard Initial work reviewed by Combat XXI Combat XXI enhancements merged into Army Modeling and Simulation Office (AMSO) Standards Nomination and Approval Process (SNAP) submission SNAP submission serving as starting point for OneSAF Objective System (OOS)
MSDL Schema Defined using XML DTD and XML Schema Sample scenarios marked up
CGF Behaviors Computer Generated Forces (CGF) systems simulate units and platforms CGF systems operate at a variety of fidelity and resolution levels Behaviors historically “hard coded” Newer systems represent behaviors in data Considerable resources required to develop CGF systems and their associated behaviors
Current CGF Systems WARSIM CCTT ModSAF Ada Finite State Machines Asynchronous Augmented Finite State Machines (translated into “C” code) Behavioral Description Frames Fundamental Behaviors (C++ code) Today’s CGF systems are built in a stove-piped manner
“Perfect World” Behavior Sharing System #1 System #2 Behaviors Represented in Common Language Using XML System #3 System #4 In a perfect world, we’d all speak the same language
CGF Behavioral Representation Logical Data Model Components Behavior Specification Logical Data Model Components Knowledge Representation Constructs Complexity Metadata Declarative Procedural Strategic
SIMTECH Demonstration Equipment Characteristics and Performance Data in XML AMSAA NGIC OTB Combat XXI
Explicit vs. Metamodel Schemas XML DTD and/or XML Schema design often involves decisions over the use of explicit tags or representation of names in data <maxspeed units=“mph”>45</maxspeed> vs. <parameter> <name>maxspeed</name> <units>mph</units> <value>45</value> </parameter>
Semantic Problems for XML Synonymy and polysemy <PERSON> vs. <INDIVIDUAL> is <SPIDER> an arachnid or software? Structural differences <PERSON><NAME>Lee Lacy</NAME><PERSON> vs. <PERSON><NAME> <FNAME>Lee</FNAME> <LNAME>Lacy</LNAME> </NAME></PERSON> Based on Hendler, 2001
Resource Description Framework (RDF) RDF data consists of nodes and attached attribute/value pairs Nodes can be any web resources Attributes are named properties of nodes Values are either atomic (text strings, numbers, etc.) or other resources or metadata instances Supports labeled directed graphs XML used as graph serialization syntax for storing and communicating RDF instances Provides basic ontological primitives Classes and relations (properties) Class (and property) hierarchy RDF triples assert facts about resources
Why RDF Is Not Enough Expressive inadequacy Only range/domain constraints (on properties) No properties of properties (unique, transitive, inverse etc.) No equivalence, disjointness, coverings etc. No necessary and sufficient conditions (for class membership) Poorly (un) defined semantics
Semantic Web (SW) Provides agent-readable descriptions of data, information, and knowledge Built on top of XML and RDF Envisioned by Tim Berners-Lee and researched by DARPA team and others W3C is beginning a Semantic Web initiative Used to define ontologies and associated instance data Huge potential for Modeling and Simulation community
Disjointness, Inverse, part-of… What is an Ontology? Thesauri “narrower term” relation Frames (properties) Formal is-a General Logical constraints Catalog/ ID Informal is-a Formal instance Disjointness, Inverse, part-of… Terms/ glossary Value Restrs. TAXONOMY ONTOLOGY Based on McGuinness, 2001: http://www.daml.org/2001/06/swday-ontologies/Ontologies-talk-060401_files/frame.htm
Beyond XML:Agent Semantics DARPA developing an Agent Markup Language (DAML) A “semantic” language that ties the information on a page to machine readable semantics (ontology)
DARPA DAML Research Team BBN Booz-Allen and Hamilton Cycorp Dynamics Research Corporation (DRC) GRCI Lockheed Martin Management and Data Systems SRI Teknowledge Carnegie Melon University University of Southern California Information Sciences Institute MIT (W3C) Stanford University UMBC University of West Florida Yale University
Benefits Standard representation of “object-oriented” concepts across the web Extensibility of ontologies through namespaces Support for complex queries involving “semantic joins” over multiple data sets
A new/old model of DoD partnering Semantic Web Res. (EU) W3C DAML www.semanticweb.org www.daml.org www.w3.org/RDF/ C2 link RDF XML RDF-S DAML-ONT DAML-LOGIC US/EU Joint Efforts (S. Decker, Coord) Horus Research efforts: SHOE OIL EC OntoWeb Intl Workshops Tools Lang Spec Demos Ctr for Army Lessons Learned EU W3C Members/directors (Dan Brickley, coord) DARPA: Funds a new generation of www technology Works closely with W3C to create a web standard Works closely with EU on international acceptance Brings DoD users (J2,J3,J6) in as early adopters Based on Hendler, 2001
Layered Architecture DAML+OIL DC PICS XHTML SMIL RDF(S) HTML XML(S) Based on Horrocks, 2001: http://www.cs.man.ac.uk/~horrocks/Slides/
DAML Status DAML+OIL ontology language released on World Wide Web Annotated “walkthrough” Examples Full definition RDFS = Resource Description Framework Schema Provides 100% mapping to XML Open discussion group run by W3C: www-rdf-logic@w3.prg Denotational (and axiomatic) Semantics published First formal semantics for a web language Proposal to W3C for standardization ongoing
www.DAML.org Language Specifications DAML Newsletter (you can subscribe) Collection of web tools Ontology library 157 ontologies as of July 9, 2001 DAML crawler over 14,000 pages w/2,000,000+ DAML statements, 5/15/01 Web tracking software used for baselining DAML use Over 500,000 hits in first 6 months
DAML Military Examples Army CALL Thesaurus Army Equipment (based on WARSIM Equipment Knowledge Acquisition Tool schema) Military task lists (e.g., UJTL) – under development Army CALL University After Next (UAN) Warrior Knowledge Network (WKN) Human Intelligence (HUMINT) interrogation procedure representation – just started Intelink (DIA) HUMINT report representation – small sample developed – currently extending
CALL Thesaurus Ontology Term name CALL Term descriptorFor RT domain subClassOf inverseOf subPropertyOf range & domain Legand preferred TermFor BT NT ACK AF entry TermFor USE UF
Military Equipment Ontology Army Military Equipment Characteristic & Performance (C&P) Ontology Provides framework to compose Army equipment from the System Unit to individual component (i.e., radio) level Ontology is modeled after data model used by WARSIM and that housed by the FDB (Functional Description of the Battlespace)
DAML Query Demonstration Semantic web DAML Demonstration Uses Call Thesaurus and Military Equipment C&P ontologies and data to demonstrate the potential of the semantic web Demonstration is composed of: Thesaurus Lookup Term Military Equipment Lookup Complex query involving multiple ontologies by using thesaurus lookup results to find military equipment terms
DAML Query Demonstration
HUMINT Report Representation Unclassified Force Protection Sample HUMINT report provided Classes identified included: Organization, Person (who) Activity, Event (what) Timeframe (when) Location Area (where) Conclusion (why)
Intelligence Report Ontology Event Location Area Timeframe Activity Equipment Organization Person Description Note Intelligence Report Conclusion Building Metadata Terrorist subclassOf partOf describedAs locatedIn within memberOf Vehicle Motorcycle Automobile drivenBy Country resultedIn conclusion metadata event conclusionEvent Subject, directObject, indirectObject, source Subject directObject indirectObject regarding source timeframe INNOVATIVE SOLUTIONS THROUGH PEOPLE, PROCESSES, AND TECHNOLOGY
Ontology Design Issue Explicit description of classes vs. encoding of information as data Explicit definition: Requires extensive maintenance of ontology Supports complex queries Encoding: Requires companion encoding standard (e.g., DIS enumeration document, SEDRIS environmental data coding specification) Decision will be made on case-by-case basis with sponsor’s goals paramount but will affect usefulness of cross-ontology “joins”
Potential Encoding Sources Events: Fineberg Verb Taxonomy Kansas Event Data System (KEDS) World Events Interaction Survey (WEIS) Codes BML operations / tasks UJTL / service task lists Things: FDMS taxonomy DIS Enumeration Document
Potential M&S Applications of Semantic Web Technologies Human Behavior Representation (e.g., CHRIS) Glossaries / Taxonomies / Thesauri (e.g., CSS) CGF Behavior Development (e.g., TRAC-M research effort) Knowledge Acquisition (e.g., FDMS) METL development (e.g., JTIMS) Scenario Development (e.g., MSDL) Unit Order of Battle (e.g., UOB DAT) Data Provisioning (e.g., AMSO SIMTECH effort) HLA FEDEP Tool Architecture (e.g., DIFs) Web Services (e.g., fly-out model responsibility) AAR / logger data
CSS Thoughts DMSO CSS provides definitions, unlike Army CALL thesaurus DAML thesaurus ontology based on ANSI standard for representing thesauri information CSS validated and traceable to Authoritative Data Sources (ADSs) DAML thesauri ontology could be extended to support definitions and ADS traceability CSS could be “ported” to semantic web instance data Once represented as SW content, terms can be manipulated using standard SW tools and other M&S-related SW applications can link to terms
Battle Management Language Standard verbiage or vocabulary used by simulation programmers or workstation controllers to describe planning and executing military functions currently performed by human controllers Key factors supporting automated Course of Action (COA) analysis: Unit identification (who) Operation / Task (what) Operation time (when) Operation location (where) Operation purpose / mission (why)
BML Thoughts BML User’s Guide Annex A contains CSS-like definitions Value of BML is in knowledge representation structure developed (i.e., units, timeframe, etc.) and specified vocabulary (i.e., User’s Guide Annex A) BML KR structure could be used as the basis for an ontology for orders BML vocabulary could be migrated into ontology instance data similarly to CSS
Preliminary BML Order Ontology Timeframe Operation Formation Task Unit partOf Location takesPlaceAt layedOutIn takesPlaceWithin arrangedAt supports performs
Summary Web technology continues to evolve XML is being used for M&S applications Semantic Web technologies emerging that provide potential solutions to DoD M&S challenges Research and proof-of-concept demonstrations needed to show benefits
Questions? Lee Lacy LLacy@DRC.com