Ontology for Enterprise Architecture Presented to Mitre EA

Ontology for Enterprise Architecture Presented to: Mitre EA Conference By: Con Kenney & Irene Polikoff Date: September 7, 2006 Federal Aviation Administration

Enterprise Architecture processes Budget Planning Cycle 1) Strategic Planning - establishes vision, mission, goals, objectives, & enterprise performance measures Direction Enterprise Architecture Business Case Development Capital Planning Portfolio Management Program Management Performance Measurement Change Operations Roadmap for Enterprise Architecture 28 February 2006 Feedback / Results Strategic Planning 2) Enterprise Architecture – Develops modernization blueprints to meet strategic goals 3) Business Case Development – Creates the OMB 300 s from selected blueprint initiatives 4) Capital Planning – Selects which initiative business cases will be funded 5) Investment/Portfolio Management – Compares investment options and guides decision making 6) Program Management – Implements the funding decisions to change the operational environment according to the modernization blueprint 7) Performance Measurement – Measure execution performance against targets Federal Aviation Administration 2

Enterprise Architecture Layers Federal Investment Management Model Has Multiple Layers DOT EA Repository, Metis, AND Portfolio Management Software Roadmap for Enterprise Architecture 28 February 2006 Federal Aviation Administration 3

EA Domain Models Roadmap for Enterprise Architecture 28 February 2006 Federal Aviation Administration 4

Two FAA Enterprise Architecture parts for NAS and Non-NAS FAA Enterprise Architecture NAS Tech Ops Non NAS Sys Ops Plan Safety Flight Serv ATO Terminal En Route & Ocean ARP ABA ARC Comm Fin ASH AVS AIO ACR AHR API AGC AGI Acq & Biz Roadmap for Enterprise Architecture 28 February 2006 AST AEP AOC Federal Aviation Administration 5

FAA Situation • Different types of applications – Air Traffic Control: Command & control • • Realtime Safety-of-life critical Multi-decade system lifespans >$150 M lifecycle – Other FAA: Business information • Online and batch • 3 to 5 year lifespans • <$10 M lifecycle • Two separate architecture repositories using different frameworks – DODAF (NAS database) – FEAF (EA Portal database) Roadmap for Enterprise Architecture 28 February 2006 Federal Aviation Administration 6

Proof of Concept for Ontology based integration • Using Top. Braid Composer automatically converted database schemas into OWL ontologies – – Tables converted into classes Columns converted into datatype properties Foreign keys converted into object properties These ‘proxy’ ontologies are used to federate queries to the underlying databases • Used existing models (FEA-RMO, TQ EA ontologies, W 3 C Geospatial ontology) as a basis for the DOTEA ontology model • Examined EA Portal database schema and NAS database schema to identify any concepts missing in the base models – Extended the DOT EA ontology with new classes and properties • When possible connected classes and properties of the proxy ontologies with the DOT EA model Roadmap for Enterprise Architecture 28 February 2006 Federal Aviation Administration 7

A concept of ‘Facility’ (a. k. a. location) in the NAS database schema Roadmap for Enterprise Architecture 28 February 2006 Federal Aviation Administration 8

A concept of ‘Location’ in the EA Portal database schema Roadmap for Enterprise Architecture 28 February 2006 Federal Aviation Administration 9

Some questions we had to consider • Are these the same concepts? • Are they different concepts? • If they are different what is the nature of their differences? • If they are the same how can we have a single point of access to the information stored in these databases? • If they are different, but complimentary, how can we aggregate the associated information? Roadmap for Enterprise Architecture 28 February 2006 Federal Aviation Administration 10

Important Advantages of Ontology Models • Documentation is part of the model – rdfs: comment and other annotation properties • Modular components – support for imports of ontologies and reference of concepts from different ontologies – universal ids and namespaces – re-use of the standard vocabularies – FEA-RMO, Geospatial ontologies, Dublin Core, … • Support inferencing to produce information not explicitly represented – inverse, transitivity, … – if a facility is located in New Jersey and New Jersey is located in the North East, facility is located in the North East Roadmap for Enterprise Architecture 28 February 2006 Federal Aviation Administration 11

EA as Layered Set of Models • Each model describes a bounded domain following rules that define – Syntax – Semantics • Each model manages a set of facts • Each model makes a set of truth claims based on its facts Roadmap for Enterprise Architecture 28 February 2006 Federal Aviation Administration 12

‘Location’ in the DOT EA Ontology Roadmap for Enterprise Architecture 28 February 2006 Federal Aviation Administration 13

The DOT EA Ontology is a collection of linked Models Roadmap for Enterprise Architecture 28 February 2006 Federal Aviation Administration 14

How do you coordinate truth claims across models? • Single, inclusive model – Conceptually simple – Complicated to implement – Hard to change • Reconciliation across models – Well-understood control – Ongoing human effort – Less flexibility for individual models • Semantic technology – More accessible but still specialized – Machine-based reasoning – Accommodates changes in models easily Roadmap for Enterprise Architecture 28 February 2006 Federal Aviation Administration 15

Bridging models through rsdf: sub. Class. Of Axioms Roadmap for Enterprise Architecture 28 February 2006 Federal Aviation Administration 16

Bridging models through rsdf: sub. Property. Of Axioms Roadmap for Enterprise Architecture 28 February 2006 Federal Aviation Administration 17

More complex mappings are also possible • Using additional OWL axioms, SWRL rules or SPAQL Construct statements – as shown in the example below This approach will be needed for the DOT EA information: • In the NAS database, State information stored in its own table • In the EA Portal database, State is just a column in a Location table Roadmap for Enterprise Architecture 28 February 2006 Federal Aviation Administration 18

SPARQL queries across the EA information repositories A single SPARQL query in Top. Braid Composer will fetch information about any locations in New York stored in either of the databases More information on using Semantic Web standards for database integration at: http: //composing-the-semantic-web. blogspot. com/ and www. topbraidcomposer. com Roadmap for Enterprise Architecture 28 February 2006 Federal Aviation Administration 19

Next Steps • Review with FAA stakeholders • Identify some capability cases and go through a tabletop exercise with the ontology • Work with Rick Murphy at GSA to make the ontology available on CORE. gov Roadmap for Enterprise Architecture 28 February 2006 Federal Aviation Administration 20