[ontoiop-forum] ontoiop_20150316: Chat Transcript

[Till Mossakowski]

*Chat transcript from room: ontoiop_20150316*
*2015-03-16 GMT-08:00*
*[09:06] **TillMossakowski: *please find the slides at 
http://iws.cs.uni-magdeburg.de/~mossakow/Slides-OntoIOp.pdf
*[09:07] *anonymous morphed into Conrad
*[09:08] **TerryLongstreth: *Till - if you're talking, I can't hear you
*[09:09] **Alexander Knapp: *Same with me ;-)
*[09:14] **TillMossakowski: *Conrad: please use UML 2.5, the relation 
between properties, compositions, attributes and associations is clearer 
there.
*[09:14] **TillMossakowski: *Conrad: UML 2.5. is essentially stable
*[09:15] **TillMossakowski: *Conrad: a property being composite means 1) 
instances cannot be owned by more than one owner, 2)...
*[09:16] **TillMossakowski: *Alexander: we need to be able to track the 
owner, if we have the instance
*[09:17] **TillMossakowski: *Conrad: properties by themselves can be 
composites
*[09:17] **TillMossakowski: *Alexander: composite properties can only 
take part in binary associations (and only in one end)
*[09:18] **TillMossakowski: *Conrad: not both association ends can be 
composite, and there cannot by cycles at runtime
*[09:22] **TillMossakowski: *Conrad: if member end that is not navigable 
is composite, things a bit hacky
*[09:22] **TillMossakowski: *Alexander will changes signatures 
accordings 2.5
*[09:23] **Conrad: *http://www.omg.org/spec/UML/2.5/Beta2/
*[09:25] **TillMossakowski: *Conrad: in uniqueness assumptions, member 
end names and attribute names are separated - but they are all 
properties. Alexander: this is a mistake
*[09:27] **TillMossakowski: *Conrad: if they are one separate ends, two 
properties can have the same name
*[09:28] **TillMossakowski: *Conrad: if they (e.g. both named a) should 
be the same property, you would have to subtype both end classes with 
the same superclass, the latter having the same property a
*[09:43] **TillMossakowski: *p.25 typo c'
*[09:44] **TillMossakowski: *what is the use of \tau's in parameters of 
query operations?
*[09:47] **TillMossakowski: *Alexander: it is possible to specify 
multiplicities of operation parameters (but this is not possible in our 
semantics currently). However, the whole issue is out of scope, because 
we do not cover state.
*[09:49] **TillMossakowski: *but we could let the operation be partial, 
and it is simply undefined in cases where the multiplicity constraints 
are not met
*[09:51] **TillMossakowski: *Alexander: we could add sentences with 
multiplicities for operations. Does the satisfaction condition hold? 
Till: yes.
*[09:52] **TillMossakowski: *(this was about slide 26)
*[09:53] **TillMossakowski: *p.28 a is not a predicate
*[09:54] *anonymous morphed into Conrad1
*[10:02] **TerryLongstreth: *Very Impressive! I hope OMG appreciates the 
work you all have done here.
*[10:07] **TillMossakowski: *sample translation of Alexander's class 
diagram to Common Logic: http://iws.cs.uni-magdeburg.de/~mossakow/UMLCD.het
*[10:12] **TillMossakowski: *source: 
http://iws.cs.uni-magdeburg.de/~mossakow/inconsistencies.uml
*[10:13] **FabianNeuhaus: *update from elisa: "The OntoIOp discussion is 
on our agenda first thing after announcements, etc. on Thursday morning, 
currently 9:30 on the agenda, although we may start it a few minutes 
early and extend into the coffee break"
*[10:16] **TillMossakowski: *strictly speaking, object diagrams are not 
like ABoxes. However, the are used like ABoxes. So we could pretend that 
they indeed are. We just should not call this a model...
*[10:18] **TillMossakowski: *Fabian: we could use DOL to connect a UML 
class diagram with a database of instances (represented in, say, Common 
Logic, or an object diagrams with non-standard semantics)
*[10:23] **TillMossakowski: *next meeting March 30th, same time
*[10:24] *List of attendees: Alexander Knapp, ConradBock, FabianNeuhaus, 
TerryLongstreth, TillMossakowski
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://listserv.ovgu.de/pipermail/ontoiop-forum/attachments/20150317/ce5280e5/attachment.html>