1 / 47

CES MWG Taxonomy Focus Group Stubbing Exercise White Paper

CES MWG Taxonomy Focus Group Stubbing Exercise White Paper. Presented By: John Dickert DoD Defense Technology Information Center At the ONTACWG Organization Meeting October 5, 2005 McLean, VA. Contributors John Dickert Mike Fontaine

kwhyte
Download Presentation

CES MWG Taxonomy Focus Group Stubbing Exercise White Paper

An Image/Link below is provided (as is) to download presentation Download Policy: Content on the Website is provided to you AS IS for your information and personal use and may not be sold / licensed / shared on other websites without getting consent from its author. Content is provided to you AS IS for your information and personal use only. Download presentation by click this link. While downloading, if for some reason you are not able to download a presentation, the publisher may have deleted the file from their server. During download, if you can't get a presentation, the file might be deleted by the publisher.

E N D

Presentation Transcript


  1. CES MWG Taxonomy Focus GroupStubbing Exercise White Paper Presented By: John Dickert DoD Defense Technology Information Center At the ONTACWG Organization Meeting October 5, 2005 McLean, VA

  2. Contributors John Dickert Mike Fontaine Dr. Glenda Hayes Hank Lavender

  3. Agenda • Practical Utility for Taxonomies • Provide background information on the focus group • Why was the Taxonomy Focus Group created? • What does the Taxonomy Focus Group (TaxFG) do? • Why build a Core Taxonomy? • Provide background on the stubbing exercise • How and why we link Community of Interest (COI) taxonomies to the core. (White Paper Topic) • Converting exercise information to web ontology language (OWL). • Process used to present and maintain the taxonomy. • Where we go from here.

  4. Practical Utility for Taxonomies • Aid for organizing • Content • Services • Structural Metadata • Aid precision search via DDMS • Aid investment analysis Country.sql ISO.xsd FIPS.xsd <ddms> : <Subject>…/MyCOI.owl#AlQaida</Subject> </ddms>

  5. Why Form a Focus Group? • OSD/NII and DISA wanted to determine the feasibility of having various COI’s make their resources visible to other COI environments via a common high-level taxonomy. • OSD needed a group to research and create a limited version of the high-level (core) taxonomy to provide a “proof-of-concept” model for use in NCES via the Horizontal Fusion pilot. • DISA wanted the same group to determine ways to connect COI taxonomies to the core to make their resources discoverable to the other participating COI’s.

  6. Focus Group Objective • OSD/NII and DISA decided a taxonomy Focus Group would be convened to: • Develop an initial core taxonomy for use in organizing and labeling DoD services/resources for Horizontal Fusion (HF), Rapid Application Integration (RAI) and NCES-related Demos. • Provide a background paper on the process for connecting COI taxonomies to the core taxonomy.

  7. TaxFG Process Timeline • The taxonomy focus group (TaxFG) chartered by the CES MWG in Nov 2003. • The initial TaxFG candidates chosen from list provided by CES MWG members. • Top-level concepts for taxonomy were taken from basic Wordnet concepts that supported DoD requirements. • Focus group developed a framework for the taxonomy based on work by M. Daconta and L. Obrst. (Semantic Web.) • “Alpha” baseline taxonomy was completed by 1 Apr 04 to meet deadline for Horizontal Fusion development.

  8. TaxFG Process Timeline • Alpha was given to HF for use in Quantum Leap II. (CES MWG membership updated during process.) • Beta update began immediately. Changes were based on continued research and understanding of top-level terms. • Final beta was presented to CES MWG by mid June 2004 and posted on GIG portal. • “Stubbing” exercise began after beta was posted to GIG. • Exercise used four existing taxonomies and data models. • Army/Marine Training Doctrine (AMTD) • Command & Control Information Exchange Data Model (C2IEDM) • Defense Technical Information Center (DTIC) • Meteorological and Oceanographic Community (METOC)

  9. TaxFG Process Timeline • Exercise tested concept of connecting COI taxonomies to Core making COI resources “visible” to other COI’s. • A white paper on “How the COI taxonomies are connected to the core taxonomy” was developed and published for use by various COIs. (Paper will be posted on DoD Metadata Registry.) • The focus group disbands (March 2005) after white paper presentation.

  10. Core Taxonomy, COI Taxonomies, and DDMS DoD Core taxonomy Organization PoliticalOrganization owl:subClassOf owl:subClassOf UrCOI taxonomy MyCOI taxonomy owl:subClassOf Group TerroristOrganization owl:subClassOf owl:subClassOf owl:sameAs ForeignTerroristOrganization TerroristGroup rdf:type rdf:type owl:sameAs al-Qaeda AlQaida Producer View Consumer View <ddms> : <Subject>…/UrCOI.owl#TerroristGroup</Subject> </ddms> <ddms> : <Subject>…/MyCOI.owl#AlQaida</Subject> </ddms>

  11. Why Taxonomies for NCES? • A central or core taxonomy was needed to help COI taxonomies: • Provide organizational hierarchy of content and services used to effectively manage, discover, and access DoD metadata information about various resources. • Help users navigate through a myriad of information and to provide more precise discovery. • Provide an avenue to “plug in” (stub) community taxonomies and expand the hierarchy to improve discoverability and usability at lower levels. • Allow inventory and monitoring services to provide metrics on content availability based on usage.

  12. Why A High Level Taxonomy for HF? • The core taxonomy was designed to: • Provide a high-level, community-neutral taxonomy in the Quantum Leap 2, Horizontal Fusion pilot. • Give users an information hierarchy built from a controlled vocabulary gleaned from DoD specific ontologies. (HF Quantum Leap 2 requirement.) • Provide the focus group a mechanism to determine “best fit” practices for connecting a limited number of COI taxonomies to the “Core”. • Establish a central taxonomy that can be expanded for all DoD in a follow-on effort and used as the baseline taxonomy for CES.

  13. Work Group Issues • Developing group consensus on foundational concepts • What is a Taxonomy? (collection, controlled vocabulary, thesaurus, model, ontology) • Taxonomy purpose? (Current objective future objective) • Scope (Core DoD, Vendor Community, Foreign partners, etc.) • Human Human vs Human Machine vs Machine Machine • Representation Standards (XML, RDF, OWL, etc) • Core Taxonomy structure (Narrow & Deep vs Broad & Shallow) • Team Composition Considerations • Professional Background and Experience • IT expertise vs functional expertise • Breadth and depth of knowledge • Commitment to Project • Availability • Participation

  14. Taxonomy Structure • Hierarchical • Multiple Independent Trees • Class-Subclass (Supertype-Subtype; Whole-Part) • Decomposition- Relatively high level (At least 3, not more than 4) • 14 Core Concepts (Culled from a first cut of approximately 24) 0 Implicit Root Node DoD Domain (Context) Core Concepts (Abstracts) 1.0 1.1 1.2 1.3 1.4

  15. DoD Core Taxonomy Taxonomic Hub Ontology COI Taxonomy COI Taxonomy COI Taxonomy COI Taxonomy COI Taxonomy COI Taxonomy Implicit Root Node DoD Domain (Context) COI Taxonomy COI Taxonomy Core Concepts (Abstracts) The Big Picture

  16. Development Resources CIM Data Buckets Activity Models Word Net DoD Publications CJCS Publications COI Taxonomies Dictionaries Professional Groups Data Models Open Cyc On-line Sources And Various Others Core Taxonomy Initial Abstract Elements (20): Original CIM Data Buckets (13): Action Documentation Location Purpose Activity Event Media Resource Agreement Feature OrganizationRole CapabilityGuidancePerson Situation Communication Information Property Structure Actions Facilities Location People Situations Agreements Funds Material Plans CapabilitiesGuidanceOrganizations Real Estate Core Taxonomy Final Abstract Elements (14): Account Asset GuidancePerson Environment Action Capability Interval LocationEvent Agreement Function OrganizationRole Core Taxonomy Abstract Development

  17. WordNet Example

  18. Core Taxonomy .51(Alpha)

  19. Core Taxonomy .75a (Beta)

  20. Core Taxonomy Examples Core Taxonomy Top-Level Abstracts

  21. Account plus 2nd level elements Account plus 2nd & 3rd level elements Core Taxonomy Examples “Account” Abstract Expanded

  22. Data Provider Registration

  23. Visual Discovery Tool -VisBee

  24. Stubbing Exercise Background • TaxFG’s 2nd charter item was to “determine ways to connect COI taxonomies to the core to make their resources discoverable to the other participating COI’s.” • Only option for the focus group was to take known community taxonomies and plug them into the core taxonomy. • TaxFG review list of known taxonomies/models and chose the following four: • Army/Marine Training Doctrine (AMTD) • Command & Control Information Exchange Data Model (C2IEDM) • Defense Technical Information Center (DTIC) • Meteorological and Oceanographic Community (METOC)

  25. Stubbing Exercise Background • Group decided on a list of process rules for the exercise and began “stubbing” COI taxonomies to the core. • The following process rules were used in the exercise: • The focus group will only use taxonomies adopted by the entire focus group. • Each team will be assigned a taxonomy. The exercise is successful if the team finds a minimum of three connections between the COI and Core taxonomy. • When connecting two taxonomy elements, the team will stub the highest COI element to the lowest appropriate Core element. (The relationship should only be applied in one direction for consistency.) • COI elements can be connected to multiple core elements.

  26. Stubbing Exercise Background • The final efforts from each team were represented in a standard spreadsheet format. • Both the final spreadsheet and comments about the process were provided to the whitepaper development team. • A white paper on “How the COI taxonomies are connected to the core taxonomy” was developed by TaxFG and will be posted on DoD Metadata Registry for use by various COIs.

  27. DTIC Use Case • Linkage based on semantics • DTIC Taxonomy, referred to as Field and Groups is officially the DoD Categorization Guide for Defense Science and Technology and available at http://www.dtic.mil/dtic/subcatguide • Preferred linkage through groups to fields and then to core • Linkage from field terms to core when appropriate • Multiple linkages possible but not common • Most common link to the Core is Scientific and Technological Research and Innovation Function

  28. DTIC Stubbing • Examples • DTIC Group Physical Chemistry (0704) subclassOf DTIC Field Chemistry (0700) then DTIC Field Chemistry (0700) subclassOf Core Scientific and Technological Research and Innovation Function • DTIC Group Aircraft (0103) subclassOf Core Equipment • DTIC Field Aviation Technology (0100) subclassOf CS&TR&I Function

  29. DTIC Stubbing • Examples (Continued) • DTIC Group Military Aircraft Operations (0102) subclassOf Core,Force Application Action and subclassOf DTIC,Aviation technology

  30. Converting the Raw Spreadsheet to OWL • Excel Spreadsheet Template created to standardize collection of taxonomy information • Visual Basic Application created to read spreadsheet and output in OWL syntax • Taxonomy (in OWL syntax) adorned with resource metadata (DDMS) required by Taxonomy Gallery • Resulting OWL document • Validated (per OWL syntax) using http://phoebus.cs.man.ac.uk:9999/OWL/TestValidator • Validated (for registration to Taxonomy Gallery) using http://diides.ncr.disa.mil/taxgal/user/submit/taxSubmitForm.cfm? • Registered in DoD Taxonomy Gallery – with CES Metadata WG approval as Core_Tax_0.75c.owl

  31. Predicate Subject Object Sometimes referred to as: Resource Property Value OWL (RDF) Version of Core Taxonomy RDF Model (Triples) Relationship Node Or: Node • </owl:Class> • -<owl:Class rdf:ID="_3" rdfs:label="Agreement"> • <owl:comment>Compatibility of observations; the thing arranged or agreed to; the statement (oral or written) of an exchange of promises; harmony of people's opinions or actions or characters. {Source: WordNet}</owl:comment> • </owl:Class> • -<owl:Class rdf:ID="_3.1" rdfs:label="Contract"> • <rdfs:subClassOfrdf:resource="#_3" /> • <owl:comment>A binding agreement between two or more persons that is enforceable by law. {Source: WordNet}</owl:comment> • </owl:Class> • -<owl:Class rdf:ID="_3.2" rdfs:label="Novation"> • <rdfs:subClassOfrdf:resource="#_3" /> • <owl:comment>The replacement of one obligation by another by mutual agreement of both parties; usually the replacement of one of the original parties to a contract with the consent of the remaining party. {Source: WordNet}</owl:comment> • </owl:Class> • -<owl:Class rdf:ID="_3.3" rdfs:label="Warranty"> • <rdfs:subClassOfrdf:resource="#_3" /> • <owl:comment>A promise about a product made by either a manufacturer or a seller. A statement or agreement by a seller of property which is a part of the contract of sale. (Lectric Law Lexicon) {Source: Lectric Law Lexicon}</owl:comment> • </owl:Class>

  32. DoD Core Taxonomy Product of CES Metadata Working Group (Taxonomy Focus Group) v. 0.75.c Workspace: https://taxonomy-fg.geden.org/ Now registered in DoD Metadata Registry

  33. Proposed Continued Plan of Action • Final work on top-level taxonomy (version 1.0) done by larger set of DoD resources in late FY05 and published for use by DoD/IC customers by early FY06. • Current focus group membership becomes nucleus used to seed DoD focus group. • Participants: Service/Agency engineers, taxonomy SMEs from Army; Navy; Air Force; Marines; HF/DIA; BMMP; DFAS; COCOMS; DISA.

  34. Focus Group Representative • Each organization within the DoD provides a candidate for the focus group. • Each focus group candidate must accept requirement to support the following: • Have and use their actual experience in the field building and implementing taxonomies to complete the focus group mission • Have the authorization to speak for and represent the interests of that organization • Provide updates on the focus group’s efforts to their respective organization • Be available for all focus group meetings • Be willing to do any homework assigned at each meeting

  35. Who will use the Taxonomy? • Who will use and how? • Developers will use taxonomy(s) to identify and retrieve informational metadata about available GIG Enterprise resources. • Service Providers (including developers) will use taxonomy(s) to advertise the availability and nature of GIG Enterprise Services. • Producers registering resources will use taxonomy(s) to advertise the availability of resources on the GIG.

  36. Who will use the Taxonomy? • Who will use and how? • Consumers will use taxonomy(s) to identify and retrieve metadata about available resources. • Developers will retrieve taxonomy(s) from the DoD Metadata Registry for reuse in Catalogs, Directories, Registries, Repositories and other software products.

  37. How to get the Taxonomy • How users access CES taxonomic information? • Taxonomy package is available for reference and reuse from the DoD Metadata Registry. (Includes amplifying information.) • Taxonomic information will also be available to Edge Users through pick lists etc. to control vocabularies when populating input forms. • Who will be Configuration Manager of CES Taxonomies? • DISA’s CES Team will provide registration procedures for COI Taxonomies in the Enterprise namespace and will manage changes to them through a CM process that is patterned after the DDMS Schema CM.

  38. Taxonomy Gallery • Purpose • Provide visibility to manage and consume taxonomies • Enable COI oversight • Reuse • Harmonization • Support Discovery • Enable navigation within and among taxonomies • Even when producer and consumer do not share common taxonomy • Ensure availability • Persistent, immutable URL lowers risk of broken links • Technical Details • Implements W3C OWL (XML) as syntax for taxonomies • Each taxonomy is self-describing (registers via embedded DDMS) Includes DoD Core Taxonomy! http://metadata.dod.mil

  39. Points of Contact • Taxonomy Focus Group Information: • Michael Fontaine – general information, taxonomy design (mfontaine@mcdonaldbradley.com) • John Dickert –Taxonomic structure, stubbing exercise processes. (jdickert@dtic.mil) • Hank Lavender – Taxonomic content, element definition structure(Henry.Lavender@hq.dla.mil) • Dr. Glenda Hayes – Data conversion, W3C standards, Web Ontology Language (OWL) (ghayes@mitre.org) • Rebecca Smith – Horizontal Fusion Integration (bsmith@mcdonaldbradley.com)

  40. How to get the Taxonomy Backup Slides

  41. Rules for Building the Core Taxonomy • Review all those elements within or accessible to the organization. (Within reason.) • Select a current taxonomy with the best fit to be a starting point. • Select Top terms that cover all knowledge for that topic. • Pick Top terms with definable boundaries to minimize overlap. • Keep the number of top terms between 15 and 25. • Keep suggested depth of thesaurus for top terms between two and four levels. • Keep suggested maximum size of a taxonomy to 256 elements. (16X16) • Understand algorithms #5,6,& 7 are experiential suggestions. • Have subject matter experts review/be a part of the taxonomy development.

  42. Rules for Building the Core Taxonomy • Use DoD strategic data buckets from old 8320 process to start process and added terrain and position to meet the requirements of items #3 & #4. (Provided reasonable representation of DoD requirements.) • Ensure the completed core taxonomy covers subjects such as: Financing, Equipping, Organizing, Manning and Directing our military forces. (Include information on terrain and opposition forces.)

  43. Core Taxonomy Roadmap • Alpha Release to HF – 4/1/2004 • Strawman Core Taxonomy – 2/24/2004 • Final “Alpha” Core Taxonomy – 4/1/2004 • Register “Alpha” in XML Registry – 4/30/2004 • Beta Core Taxonomy – 6/30/2004 • Provide Core Taxonomy Business Rules– 5/7/2004 • Editing and redesign of Core taxonomy – 5/8 to 6/15/2004 • Beta Core Taxonomy completed – 6/30/2004 • Beta Core Taxonomy presented to CES MWG – 7/2/2004 • Vetting Process – 7/2/2004 to 9/27/2004 • Registered in XML Registry – 09/29/2004

  44. Core Taxonomy Roadmap • Taxonomy Stubbing Exercise • Proposal from OSD to connect COI taxonomies to Core – 4/15/2004 • Initial 3 touch points business rules– 5/28/2004 • Final Stubbing Relationship Business Rules – 6/23/2004 • Final Candidate COI taxonomies List – 7/9/2004 • Stubbing Exercise – 7/20/2004 to 9/28/2004 • Compile finding to Spreadsheets – 9/29/2004 to 10/19/2004 • Draft white paper – 11/09/2004 to 01/28/2005 • Final Report w/stubbing package – 02/16/2005

  45. Key Positions • Focus Group Lead: Builds meeting agenda, approves presentation slides and provides direction for focus group, determines group assignments, supervises topic voting process, determines “rules of engagement”, finalizes voting membership list. • Technical Lead: Provides technical direction and oversight for agenda items and “homework”. Coordinates agenda with FG Lead and facilitator. Moderates homework assignments. Collects and disseminates background information and deliverables. • Facilitator: Moderates meetings, enforces agenda, coordinates focus group assignments (homework) with technical lead, moderates topic voting process, enforces “rules of engagement”, disseminates minutes, builds voting membership list. • Meeting Coordinator: Researches, coordinates, and disseminates times and locations for future meetings to focus group membership. Disseminates background information (directions, assignment POCs, presentation slides, conference call information, attendee lists etc.) to membership. • Technical writer: Collects information to build meeting minutes.

  46. Rules of Engagement • Follow the agenda: The focus group will remain “focused” by keeping to the agenda. (Sidebar discussions will be taken outside.) • One idea at a time: To accomplish each agenda item, the focus group will discuss and complete an agenda topic before moving to another tropic. (No jumping around) • One speaker at a time: When discussing an idea, one speaker will have the floor. Open discussions will be lead by a moderator. • Keep the discussions professional: Members will keep the discussions focus and professional. • Stick to the schedule: Meetings are 2 hours long with a 10 minute break after the first hour. Discussion times will fit the 2 hour limitation. • One vote per person: When topic votes are presented, voting members have one vote per topic.

  47. Ontology Spectrum: Weak to Strong Semantics Strong semantics Modal Logic First Order Logic Local Domain Theory Description Logic Is disjoint subclass of with transitivity property DAML+OIL, OWL Unified Modeling Language Conceptual Model RDF/S Is subclass of XTM Extended ER Thesaurus ER Has narrower meaning than Schema Taxonomy Is subclassification of Relational model Weak semantics Source: The Semantic Web, Michael C. Daconta, Leo J. Obrst, Kevin T. Smith

More Related