Dr. Leo ObrstDr. Leo ObrstMITRE MITRE

March 14, 2006March 14, 2006

Upper Ontology Summit Issues: Upper Ontology Summit Issues: Opportunities and ChallengesOpportunities and Challenges

Tuesday, March 14, 2006Tuesday, March 14, 2006

Upper Ontology SummitUpper Ontology SummitNISTNIST

2

Ontology Elephants

There is no single real elephant

There must be an upper elephant

An elephant is abstract

An elephant is very abstract

There must be a purpose for an elephant: use cases?

An elephant is really very simple An elephant is the

result of consensus

Open vs. Closed Elephant

There are only distributed elephants & their mappings

3

1. Points of Agreement

1. We desire semantic interoperability.2. We agree that a mere taxonomy is insufficient for that.3. We agree that axioms are an indispensable part of creating semantic

interoperability.• What types of semantic interoperability require axioms and what types could be

achieved with taxonomies with verbal definitions?

4. We agree to disagree on the best way to achieve interoperability: establishing (partial) alignments/correspondence, common subset, common upper ontology, etc., as will be discussed in (II).• We agree that the best combination of mechanisms may emerge from the work to

be undertaken

5. We agree that the dialog on making upper ontologies interoperable will contribute to improvements in the existing upper ontologies.

4

2. Opportunities and Challenges - 1

1. Modularity: How to Achieve? Incompatibilities?• Lattice/poset of theories?• 3-D vs. 4-D vs. 3-D/4-D• Other?

2. Mapping among the terms/axioms of the relevant upper ontologies• Term to term maps• Bridging axioms• Finding consistent common interpretation subsets• Create a reference library of upper ontologies• Express in a common language: CLIF or some other CL dialect, IKL?

Do we need a common language for expressing commonalities and differences?• Identifying an intersecting subset of terms/axioms of the relevant upper ontologies• Identifying areas where agreement is easy vs. areas where agreement would be difficult• Barriers to agreement, e.g., assumptions on which ontologies are based

Scope of these assumptions: do two ontologies based on different assumptions have no points of agreement or are some elements neutral with respect to assumptions?

Seek agreement on what needs to defined, rather than reach agreement on definitions?

3. Criteria for evaluating ontologies, including application-based criteria• Criteria for evaluating ontologies, including application-based criteria • Methods for evaluating ontologies, including protocols and testbeds• Certification?

5

3. Next Steps?

1. Create a consortium or working group• Align with existing standards organization?• Independent: New committee under NCOR, ECOR, JCOR:

Upper Ontology Reconciliation and Mapping Committee?2. Identify (or begin to identify) pairs of upper ontologies and/or content

elements across upper ontologies where alignment work would be particularly appropriate

3. Seek joint funding to address 2 and 3.1 and 3.24. Other?