wiki:meeting101216

(This being notes from what is supposed to be a fortnightly meeting on general information modelling issues, as applied to CSML, MOLES, and our support of the MetOcean OGC working group and various INSPIRE activities.) (last meeting notes)

In this meeting we concentrated on what is needed to get MOLES 3.4 out the door.

We went through the complete list of tickets on milestone:'V3.4 UML Final'

From the point of view of those not at the meeting, the key conclusion we came to was that we should concentrate on getting moles 3.4 out the door, and that meant postponing some activities to a future version (so ticket:47 and ticket:49 got bumped - to do with MO_Computation and MO_Source - since they have dependencies on Metafor (see their  wiki page).

We discussed ticket:51 in detail: There is an issue with OM_Process and OM_Observation in that ideally we'd like the former to be reusable, which implies that it should not include any observation specific parameters. This flies in the face of what the EA notes say about OM_Process:

An instance of OM_Process is often an instrument or sensor, but may be a human observer, a simulator, or a process or algorithm applied to more primitive results used as inputs. (NOTE ISO 19115-2:2008 provides MI_Instrument, LE_Processing and LE_Algorithm, which could all be modelled as specializations of OM_Process. OGC SensorML provides a model which is suitable for many observation procedures.)

We would argue that there is a contradiction in this definition between nouns and adjectives/verbs. Either we consider a process of consisting of using something (algorithm or instrument) or it is a descriptoin of the algorithm or instrument. In the latter case the event specific things like algorithm parameters actually used, or method of using the instrument, or observation protocol, should be part of the observation itself.

The latter feels inherently correct, so that for example, a flight should not be part of the process, but part of the observation ... however, somehow, a laboratory facility feels like the right place to find information about a chemical analysis process ... and that feels right ... but they're contradictory feelings. It comes down to a choice to be made by those making the O&M profile (us). Currently (V3.4) we've put everything in the process (e.g. flight), but now it's less reusable. That only matters if we want to sort on the process per se ...

Further discussion of this issue has been postponed to V3.5 or later. Other issues discussed included the perennial ticket:15: how to do outbound links. We have settled on using CI_OnlineResource as it has more useful semantics in the UML than somehow propagating xlink.

Other discussion appears in the specific tickets (ticket:44, ticket:46, ticket:48, ticket:43, ticket:50 and ticket:50). The latter was about the utility of a moles specific observation collection class. We no longer think we need one of these since we can make a collection of observations either the result of a moles observation, or we can use the related_observation link.

 Trac Powered
Site hosted at the
British Atmospheric Data Centre
for the