AML Meeting June 2013
AML Modeling Meeting
The AML modeling meeting will be a full day on Tuesday, June 18, at the OMG meeting in Berlin, Germany. We will also try to schedule some "pre-meeting" meetings with other participants (e.g. Elisa Kendal, Michael Van Der Zel, ?) on Monday or, if not possible, Sunday afternoon. Our goal is to maximize the Tuesday Face to Face time to attempt to get enough done that work can proceed on the AML specfication.
The AML submission to the OMG is anticipated to consist of three parts:
- An AOM metamodel - a model of what we are trying to represent in the profile
- A UML profile that implements the metamodel. It is anticipated that this profile also will be based on parts of ISO 11179-3 and OMG CTS2.
- Validation rules in OCL that specify UML + profile requirements for complete and correct archetype models.
- Transformation rules (using QVT or ATL) to map between UML and AOM meta models
The questions below are some of the issues needing resolution before the AML project can continue. Our goal is to prioritize this list, marking those questions that can only be settled by in person discussion, those that require both in person and e-mail or other dialog and those that can be resolved post meeting. We will use this discussion list to advise the agenda.
- Submission contents:
- AOM Metamodel in UML
- RFP requirements - we need to give this a once over and decide what must, will and will not get done on the initial submission
- What is and is not priority (e.g. metadata? ...) for first submission
- OCL - how much do we put into the profile definition?
- Beale - statements in AOM about applicability of OCL
- ADL requirements
- AOM and Terminology Binding - should the AOM model be "fleshed out" to include a full terminology binding part as part of the metamodel, or should it be used as it currently exists, with the "fleshing out" being done in the UML Profile, with anticipation that it will subsequently be enhanced.
- What more needs to be done to create a publishable metamodel from AOM?
- AOM constraint model - does this require separate UML profile, or only specification of how UML is used?
- Coded Ordinal model - AOM has model specific ordering, HL7 represents this as part of the terminology, meaning that the ordering is independent of the recording model. What should we do here?
- What contribution can/should ISO 11179-3 make in AML metamodel and profile for terminology representation and data element binding?
- AOM metamodel as defined in openEHR has data types that are used by both the Reference Model and AOM. How do we separate an AOM from the reference model from the data types?
AML Representational Issues
- Multiple designations and languages for each model element, including archetypes and terminology, names and definitions.
- AOM slots and choice groups. Current AOM/ADL requires archetype name pattern matching for slots, which may be problematic in UML. Should this even be based on pattern matching or should the model that underlies the matching be made explicit?
- Use of attribute existence vs. occurrence vs. cardinality, and optimal representation in UML.
- Versioning of individual archetypes, vs. larger UML model that contains multiple archetypes
- Need CIMI requirements for archetype authorship and registration metadata. Can this be aligned with ISO 11179-3?
- MIF2 UML Profile leasons learned / compare (Michael)
The following background will be assumed
- ADL 1.5 and AOM 1.5 Specifications, http://www.openehr.org/programs/specification/releases/currentbaseline
- openEHR Identification System for Knowledge Artifacts, http://www.openehr.org/wiki/display/spec/Development+and+Governance+of+Knowledge+Artefacts
- UML Profiles
- ISO 11179-3 Edition 3
- Slides from CIMI Presentation Tuesday, June 11
- CTS2 Value Set Definition / Resolved Value Set and EntityDescription API's. The links below go the CTS2 1.1 Release, which is being voted on on the Berlin meeting (informative only - not needed for in-depth study):