Download presentation
Presentation is loading. Please wait.
1
DCMO DM2 Ontology Development
12/12/11
2
Perspectives on Objectives
DCIO Objectives Consistency Re-Use Modeling & Simulation Method/Tool Agnostic Architecture ‘cross-walk’ Decision Support Interface to ADS IDEAS foundation for coalition partners and mathematical precision DCMO Objectives Common Vocabulary Re-Use Business Process Focus Industry Standards Based Federated SPARQL Queries Analytics/Decision Support Interface to ADS Linking Concepts (and data) across Domain Vocabularies using OWL Those in Green are considered identical and so don’t need much discussion.
3
Perspectives on Objectives
“A standard meta model is only one component in achieving consistency; controlled vocabularies, taxonomies, and/or common reference data in the instantiated meta model are also necessary.” DCIO Objectives Consistency Re-Use Modeling & Simulation Method/Tool Agnostic Architecture ‘cross-walk’ Decision Support Interface to ADS IDEAS foundation for coalition partners and mathematical precision DCMO Objectives Common Vocabulary Re-Use Business Process Focus Industry Standards Based Federated SPARQL Queries Analytics/Decision Support Interface to ADS Linking Concepts (and data) across Domain Vocabularies using OWL DCMO focus is on E2E value chain management, standard BPMN ‘’primitives’ for consistent business process representation and ultimately process model simulation, analysis, reengineering , and execution DCMO shares the objective of tool agnosticism, and is focused on adopted business industry consensus standards spanning process, data, and rules DCMO objective is also cross walking information from systems beyond traditional architecture tool vendors that may conform to different standards and models appropriate for their particular business domain DCMO objectives also include performance metric analytics based on linkage to Authoritative Data Quotes are from DCIO “Purpose of the DoDAF Meta Model (DM2)” – 13 September 2011 “In many cases the ADS for architectural description information is not an architecture repository, but some other type of data asset.”
4
DCIO Specific Objectives for DM2
Specific DM2 Objectives Vocabulary for description and discourse about DoDAF models and views and their core process usage Basis for Physical Exchange Specification for exchange of data between architecture tools and databases Basis for semantic precision in architectural descriptions to support … integration and analysis in support of core process decision making Information Sharing across DoD EA COI using precise, universally understood, and commonly interpretable semantics.
5
DCMO DM2 OWL Approach Started with classes from the DM2 conceptual model Expanded conceptual class model using the DM2 logical model All DM2 IDEAS based constructs that could be expressed directly in OWL were replaced with standard OWL constructs Any IDEAS based constructs that were not required to express DM2 were not included Used LDM for more detail, especially in the relationships described in DM2 by subtypes of IDEAS:couple
6
Working Group Decisions & Tenets
Objective is to model or identify a complete DM2 OWL file (10/26/11) The target DM2 OWL file will support OWL-DL( prior to 10/26/11) IDEAS reified OWL file is too complex and will not be used (10/26/11) DM2 OWL will be based on DM2 Conceptual and Logical Model(11/2/11) Will not use IDEAS Foundation based DM2 Ontology (11/2/11) The IDEAS Foundation based DM2 OWL file created and uses new constructs, many of which are already represented in OWL IDEAS Foundation based DM2 OWL creates and uses some complex constructs that are not needed to model DM2; e.g. Powertype The goal is to describe DM2 completely in OWL open specification based constructs and not in IDEAS Foundation specific based constructs The new CIO DM2 Errata Ontology is OWL Full and will not be used This group will create a pure OWL based DCMO DM2 Ontology (11/2/11) Risk: If CIO develops alternate IDEAS based DM2 OWL, it may be challenging to map DCMO OWL classes and properties to the “final” CIO DM2 OWL without loss.
7
DCMO Design Decisions Removed many IDEAS foundation concepts:
Individual, Type, Thing, Property typeInstance, powerTypeInstance, superSubType tuple, couple, propertyOf IDEAS:Thing => owl:Thing IDEAS:Individual, IDEAS:Type => owl:Individual, owl:Class IDEAS:couple => owl:ObjectProperty Note: tuple, couple replaced with Object Properties using domain/range and also anonymous Restriction classes. IDEAS:Property => owl:DatatypeProperty Note: Datatype properties are used to capture measures. This may change based on our use-case analysis going forward. IDEAS:DM2Names => rdfs:label IDEAS PowerTypes largely ignored or replaced as owl:Class subclasses as applicable. IDEAS foundation concepts removed are already ‘built in’ to the OWL language. All DM2 relationships are of type IDEAS:couple
8
Next Steps Map BEA and BPMN 2.0 to DCMO DM2 OWL file
Begin testing by storing BEA data and SPARQL Querying Modify DCMO DM2 OWL as needed Analyze for possibility of replacing portions of DCMO DM2 ontology with valid horizontal ontologies; e.g. Location, Measures, etc
9
Back Up
10
DCIO DM2-errata Ontology Discussion
Places IDEAS Foundation classes on top of OWL by extending core RDF/RDFS properties Specifically: rdf:type and rdfs:subClassOf Includes IDEAS foundation and IC-ISM concepts in one OWL (DM2) file these should be broken out into separate ontologies (like they did with the XSD version) and referenced as imports to DM2 these two ontologies should be able to stand on their own separately for reuse Implies an OWL-Full Ontology Classes become Individuals Properties become Classes
11
Working Group Decisions & Tenets
“complete DM2 OWL file “ means complete representation of concepts defined by DM2 (not necessarily all IDEAS concepts) Objective is to model or identify a complete DM2 OWL file (10/26/11) The target DM2 OWL file will support OWL-DL( prior to 10/26/11) IDEAS reified OWL file is too complex and will not be used (10/26/11) DM2 OWL will be based on DM2 Conceptual and Logical Model(11/2/11) Will not use IDEAS Foundation based DM2 Ontology (11/2/11) The IDEAS Foundation based DM2 OWL file created and uses new constructs, many of which are already represented in OWL IDEAS Foundation based DM2 OWL creates and uses some complex constructs that are not needed to model DM2; e.g. Powertype The goal is to describe DM2 completely in OWL open specification based constructs and not in IDEAS Foundation specific based constructs The new CIO DM2 Errata Ontology is OWL Full and will not be used This group will create a pure OWL based DCMO DM2 Ontology (11/2/11) Risk: If CIO develops alternate IDEAS based DM2 OWL, it may be challenging to map DCMO OWL classes and properties to the CIO IDEAS OWL without loss. “OWL-DL” means capable of using DL reasoners (e.g. Pellet, Fact++) to compute logical entailments and inferred models – no clear use cases identified but don’t want to rule it out “IDEAS reified OWL file” represents IDEAS tuples as OWL classes (not object properties as relationships are typically represented in OWL), “DM2 Conceptual model is … data dictionary.” “pure OWL” means …
12
OWL Full Explanation ** Thanks Steve!
Because the ontology asserts that dm2:typeInstance is a subproperty of rdf:type, rdf:type is asserted to be an object property. Protégé places it as a subproperty of owl:topObjectProperty. The blue nodes in the picture shows are example individuals of two DM2 classes: dm2:Type and dm2:Powertype. The picture also shows an assertion of the dm2:typeInstance property between them: typeInst is a type of powertypeInst. That captures the IDEAS concept of type relationships. However, because the ontology asserts that dm2:typeInstance is a subproperty of rdf:type, the triple <typeInst dm2:typeInstance powertypeInst> can be used to infer <typeInst rdf:type powertypeInst>. In OWL, an assertion that some individual is the object of a triple with rdf:type as the predicate lets a reasoner infer that the individual is an OWL class. Therefore powertypeInst is both an individual and a class, which can only happen in OWL-Full. ** Thanks Steve!
13
[Steve’s DM2 errata email)
Gregory, That's an interesting structure. I do want to raise the issue that the ontology is (I think) OWL-Full rather than OWL-DL. Please check my logic to see if you agree. The ontology defines the following RDF graph (ignore the blue nodes for a moment): Because the ontology asserts that dm2:typeInstance is a subproperty of rdf:type, rdf:type is asserted to be an object property. Protégé places it as a subproperty of owl:topObjectProperty. The blue nodes in the picture shows are example individuals of two DM2 classes: Type and Powertype. The picture also shows an assertion of the dm2:typeInstance property between them: typeInst is a type of powertypeInst. That captures the IDEAS concept of type relationships. However, because the ontology asserts that dm2:typeInstance is a subproperty of rdf:type, the triple <typeInst dm2:typeInstance powertypeInst> can be used to infer <typeInst rdf:type powertypeInst>. In OWL, an assertion that some individual is the object of a triple with rdf:type as the predicate lets a reasoner infer that the individual is an OWL class. Therefore powertypeInst is both an individual and a class, which can only happen in OWL-Full. I ran the ontology, plus the individuals, through 3 reasoners. One (Fact++) crashes. The other two (Pellet and HermiT) accept the ontology uncomplainingly. Both infer the rdf:type triple. Neither reports that powertypeInst is a class – or at least, if they do Protégé doesn’t display that fact. In other words, either means my logic’s wrong, the reasoners aren’t prepared to make that kind of inference, or Protégé doesn’t report it. Regards, Steve Wartik
14
IDEAS – DM2 Support for Semantic Webs
“Linkability of datasets via URI’s … Works” “Managed URI’s … Works – reduces unintended redundant data and enables more accurate and complete queries” “If needed, IDEAS and DM2 can help in making the links semantically stronger since every “link” (i.e. DM2:couple) is supertyped to something mathematical” Quotes taken from DCIO presentation of same name dated 11 July 2011
15
IDEAS – DM2 Support for Semantic Webs
“OWL can be argued as more syntactical than semantic.” Question: does representing DM2 constructs in OWL syntax preserve their mathematical meanings? Quotes taken from DCIO presentation of same name dated 11 July 2011
Similar presentations
© 2024 SlidePlayer.com. Inc.
All rights reserved.