[Cesg-all] Upcoming OMG meeting regarding ODM

[Greg Kazz]

All,

Concurrent with the upcoming Fall CCSDS meetings in Atlanta, Georgia, will 
be the Object Management Group (OMG) Meetings. One of the topics that will 
be discussed at these meetings will be ODM, OMG's technical committee
working on the "Ontology Definition Metamodel" (ODM).

Nicolas Rouquette, from NASA/JPL is planning to attend and participate at 
this meeting using the CCSDS Influencing Membership. Please not that CCSDS 
has an "Influencing Membership" with the OMG, but has no voting rites. 
However, technical participation at this meeting is welcome including other 
agencies and centers besides NASA/JPL.

Below is a synopsis from Nicholas on the goals of ODM along with NASA/JPL 
concerns/issues for this meeting.
Please let me know if any other agencies wish to participate in these 
discussions at the OMG meetings, since CCSDS needs to have a coordinated 
approach.

The ODM is intended to support practitioners working with a variety of 
modeling tools such as UML, Entity-Relation schema (ER), Topic Maps (TM), 
Common Logic (CL) and the Semantic Web Ontology language (OWL). The ODM is 
expected to produce metamodels for subsets of each of these
languages such that models developped in one language can be converted into 
another language with the strong guarantee that the semantics of the 
original model will be preserved across the conversion.

The ODM is charted to align the metamodels of modeling
languages commonly used for documenting the architecture of complex
systems, such as a profile for UML. Since UML 2.0, there is renewed 
interest and effort to provide practical
support for several standards:
from AP233 to SysML, EDOC and RM-ODP. RM-ODP itself is heavily used in
several CCSDS standards and documents.
This stack leads to concerns of mismatch across standards in a similar
manner that Garlan and Shaw's famous observations
and descriptions of architecture mismatch issues. The stack of standards
involved at the level of CCSDS is very broad
and involves several layers. The ODM provides architects a new
perspective for analyzing the stack of standards used in a project.
For example, the ODM mappings across metamodels facilitate comparing the
semantic roles of the same concept
across the different views where it is used and across the different
levels of abstractions where it is described.
These comparisons are an important aspect for the practical application
of these standards.

At NASA/JPL, there is increasing recognition that many standards and
recommendations, while available, involve significant
pragmatic challenges to adopt such standards and follow recommended
practices as a tool-supported engineering
and management methodology with the assuredness that the resulting
project will be significantly better because of it than without.
While on-going work on such standards and recommendations will address
some of these challenges, there is irreducible issue of
conceptual modeling common to these standards: an architect still needs
to specify several kinds of objects and relations according
to the specific needs of a domain or project. Several projects at JPL
have recently approached this specification activity
from the semantic web perspective, either in terms of metadata (e.g.,
OODT), taxonomy (e.g., the NASA taxonomy),
web services (e.g., sciflo) and more recently, ontologies (e.g.,
SWEET).  The semantic web stack (e.g. RDF, OWL) addresses
pragmatic issues of "modeling" and "semantics" that are almost
orthognonal to the pragmatic issues of "architecture" and "organization"
for which the stack of relevant standards from ISO, CCSDS, OMG is
particularly useful. Combining these two stacks seems
to be a timely activity for JPL and NASA, particularly since increased
attention to resource management, budget oversight
and mission accomplishments will eventually require greater ontological
rigor at the levels of the engineering models used
such that the RM-ODP notions of viewpoint consistency and conformance
testing become practically measurable and analyzable.