Page tree
Skip to end of metadata
Go to start of metadata

Date

Attendees



Agenda

1) Use Case reminder

2) Where we are on our road map. 

3) Open Action Items

4) JIRA Issues Review - https://jira.edmcouncil.org/projects/FND/issues/FND-48?filter=allopenissues

5) Todays content discussion.

SMIF OWL-UML

SKOS

RDF/S

6) For next week.

Proceedings:

20180904 FIBO FND

 

 

Because the FIBO System of Record is OWL, we believe that FIBO OWL satisfies the OMG technical requirement as the normative spec. To support this spec, it is intended to ingest FIBO into a clean CCM project and to generate a new complete XMI from that.

 

FIBO is not a model driven specification. MB questions why the FIBO Project MDZIP file should be other than Informative. EK confirms it should be Ancillary, the definition for this being artifacts form which any normative part of the submission was derived. In this case, the diagrams were derived from the FIBO. Project repository (MDZIP file). 

 

Diagrams: whether we need every element to be in a diagram (EK) or if the diagrams can just be illustrations (DW). MB confirms that the current Annexes, which contain the diagrams, are asserted to be a Normative part of the specification.

 

Cory Single Class defining diagrams can be done now. Cory has the code now. These diagrams set the context of the class.

 

MagicDraw today can automatically generate a taxonomy of each diagram. MB: This is a separate diagram type supplied in MagicDraw but it would be preferable to automatically generate taxonomy diagrams in the Class Diagram format to support the other diagrams for the classes and properties.

 

Bobbin This is already in the containment tree. MB these still have to go in a diagram. By hand a set of detailed diagrams for one ontology takes about 2 hours. Many diagrams take less but we should allow 2 hours per ontology for determining how many diagrams and then creating them.

 

Elisa believes that automation cannot make human readable diagrams. Cory changes his mind completely. Now says all work must be by hand if we were to try to meet that requirement. Otherwise autogenerate diagrams can be improved upon by hand afterwards. Later automation can establish that these hand curated diagrams still cover every element.

 

Bobbin Teegarden Can we set aside a few hours at the upcoming OMG Ottawa for this discussion? MB: We can put this on the agenda – we have an agenda planning call tomorrow (Monthly Update Call).

 

MB Should the MDZIP Project file(s) be ancillary or informative. EK They should be ancillary. If all elements that are not in release are deleted from the CCM model files, then the generated XMI should be complete and match the OWL.

 

EK Need to say that the ancillary Project file corresponding to the FIBO-Master one from which the diagrams were generated might contain data that should be ignored. MB to avoid duplications in the XMI, we will regenerate that from a separate Project file into which we have ingested the submitted OWL. There will be no diagrams in this separate MDZIP project. There will then be 1 ancillary file that is a clone of FIBO Master (with the non Release stuff removed) and a second one that is not a clone from. Master but is where we have ingested the OWL, and from which we generate the XMI. FIBO Master itself (on TWC server) will still have some metadata and other content that will eventually be deleted. MB can remove some of the stuff that PR has identified as being redundant.

 

Cory Whatever produced the diagrams should also be in ancillary. EK Otherwize this could not be reproduced by others. The source of the submission and the ancillary should be the same. For future OMG submissions we should consider the proposal to start with a clean set of models that have not had the previous hand work and complex history. This would require that we re-do the diagrams but in the long run this will be less of an overhead than maintaining the current FIBO-Master project with the duplicate metadata etc. that result from its history.

 

DW This entire conversation is silly. It is like asking a class of 1st graders to draw a picture of - you name it. All will be different.

 

Bobbin Teegarden Can we see a demo of what can be done? Re Cory’s description of autogenerated diagrams, at OMG, and have this conversation after we see? Then in the future, do we do the OWL in RDF first, then draw diagrams of what they mean?

 

ACTION: Cory wrote a suggested structure and rules for diagrams. Cory thinks that he sent this to MB. MB and Cory will search for this. MB also wrote what he thinks. https://wiki.edmcouncil.org/display/FPT/Diagram+Editing+Conventions

 

Bobbin Teegarden Will this set new precedents for OMG?

 

Cory to EK There is really no OMG requirement for XMI. It is the culture. EK, True but others want to use the XMI as a starting point. Cory, it is not a true requirement. EK describes a number of specific end user examples of people who would use the XMI. MB documentation of what would be useful and who would use it is very important, please can we have that written down. EK, we have this but there were many many changes in how the submission looks now.

 

DW One proposal today was to use CCM to ingests the FIBO OWL into a fresh project and then use Cory’s automation facilities to build whatever it can do to generate a diagram. Then we tweak it from there. How do all of these thoughts support that idea?

Decisions:

Action items

1 Comment

  1. Here are the slides: