[ontoiop-forum] ontoiop_20150112: Chat Transcript

[Till Mossakowski]

*Chat transcript from room: ontoiop_20150112*
*2015-01-12 GMT-08:00*
*[08:03] **TillMossakowski: *Report from OMG meeting (Fabian)
*[08:04] **TillMossakowski: *Generally, the feedback was positiv and
encouraging. Industry people see the potential for industry.
*[08:04] **TillMossakowski: *Noone seems to have read the proposal
though. Probably this will change with the final proposal.
*[08:05] **TillMossakowski: *The only critical point is the integration
of UML class diagrams and its integration with other languages. Various
useful scenarios have been discussed. UML class diagrams are extrmely
important to OMG.
*[08:06] **TillMossakowski: *The problem is that UML class diagrams do
not have a formal model-theoretic semantics.
*[08:06] **TerryLongstreth: *I've just left a meeting with another
standards group, and one of the other attendees is from the Planetary
Data System (PDS) project at JPL. He was much interested in OntoIop. Can
I share one of our drafts with him, or can one of you contact him about it?
*[08:07] **TillMossakowski: *Tne OMG people did not like ignoring the
issue, nor using the model-theoretic semantics by Alexander Knapp. The
do not want to have different semantics for UML. There is already fUML.
*[08:08] **TillMossakowski: *However, fUML is an operational semantics,
while for DOL, we need a model-theoretic semantics.
*[08:09] **TillMossakowski: *Michael: are OMG people open for extending
the fUML axiomatization (in Common Logic)?
*[08:09] **TillMossakowski: *Fabian: there is no mapping from UML class
diagrams to CL axioms.
*[08:09] **TillMossakowski: *We would need to come up with such a mapping.
*[08:11] **TillMossakowski: *Fabian convinced them that we need a
model-theoretic semantics- Several fUML people in the room offered their
help.
*[08:11] **TillMossakowski: *We should not take something from the
literature, but base the work on fUML.
*[08:12] **TillMossakowski: *Open questions: how would the collaboration
work? Where/how is the resulting document to be published?
*[08:14] **TillMossakowski: *An easy way of providing a mechanism for
OMG people to check our translation from UML class diagrams to CL would
be to implement it (say in Hets), such that they can make experiments.
*[08:14] **TillMossakowski: *This would be a better way for them,
compared with providing an institution.
*[08:17] **TillMossakowski: *And we would get an institution by
"borrowing" model theory from CL via the translation.
*[08:18] **TillMossakowski: *Rick: How does OntoIop plan to address
parametricty available through UML and not available in fUML?
*[08:18] **rick: *I need to check my audio setting
*[08:20] **TillMossakowski: *Rick: in section 17 of the UML standard you
find templates. They are not used so frequently.
*[08:23] **TillMossakowski: *templates could be handled on the level of
theory morphisms (or generic specifications provided in CASL). That is,
they are treated not as part of institution/logic, but as part of the
structuring mechanisms on top of institutions.
*[08:24] **TillMossakowski: *Rick: treating parametricity in
higher-order logic has some benefits
*[08:25] **TillMossakowski: *Fabian: strategically, it makes sense to
stick with the fUML subset of UML. If we want to put something in that
is outside fUML, we will re-open their discussions, causing delay.
*[08:26] **TillMossakowski: *Fabian: OMG people are aware that UML is a
complex multi-headed beast - not even the exact boundaries of UML class
diagrams are clear.
*[08:28] **TillMossakowski: *Rick: has done some work on functional
abstraction that could be useful for OntoIOp.
*[08:29] *anonymous morphed into ConradBock
*[08:31] **TillMossakowski: *we should focus on producing a draft of the
translation of UML class diagrams to Common Logic as soon as possible,
so that we can provide it for the OMG meeting mid/end of march (which
means that we need to submit mid February).
*[08:34] **TillMossakowski: *Conrad: we could simplify the fUML
axiomatisation by omitting the aspect of time, which is not needed for
the axiomatization of classes.
*[08:35] **TillMossakowski: *Michael: the PSL ontology can be used (as
it is in fUML) for capturing behaviour.
*[08:39] **TillMossakowski: *Conrad: OMG people probably won't mind, as
long as the result is consistent with fUML.
*[08:40] **TillMossakowski: *Note that we *only* promise to cover *UML
class diagrams*.
*[08:47] **TillMossakowski: *Ed Seidewitz's paper also contains meta
predicates like class(C) ("c is a class"). Do we want them in?
*[08:48] **TillMossakowski: *It seems that we need these, e.g. for
typing purposes.
*[08:49] **TillMossakowski: *Different topic: what is the status of the
translation paper? Appendix to the DOL standard? to fUML?
*[08:49] **TillMossakowski: *Fabian: OMG people say that this needs to
be discussed. For now, it can be an informative annex. In the long term,
it might migrate.
*[08:50] **TillMossakowski: *Fabian: we don't want to wait until there
is an update of fUML.
*[08:54] **TillMossakowski: *We decide to develop the translation of UML
class diagrams into Common Logic as an informative annex of the DOL
standard. Deadline is February 23rd (four weeks before the Reston OMG
meeting starts).
*[08:56] **TillMossakowski: *Fabian: at the Reston OMG meeting, we
should organise a meeting with the revision task force (RTF).
*[08:59] *anonymous morphed into JimLogan
*[09:01] **TillMossakowski: *The DOL standard draft is freely available
at https://github.com/tillmo/DOL Feel free to share it with anyone.
*[09:01] **TillMossakowski: *next meeting: January 26

*List of attendees: ChristophLange, ConradBock, FabianNeuhaus, JimLogan, MichaelGruninger, OliverKutz, TerryLongstreth, TillMossakowski, RickMurphy, MihaiCodescu*

-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://listserv.ovgu.de/pipermail/ontoiop-forum/attachments/20150112/21e22b92/attachment.html>