Devices Ontology Working Group Meeting 2008.02.19

A. Review of Instructions for Providing Device Data

The instructions were not ready, but we reviewed the issues with the old instructions (overview and template instructions). Two primary issues are (a) 'one line per instrument' isn't very user-friendly, and (b) without concrete vocabularies to refer to, people get confused about vocabulary terms.

There is also the element of context -- in different contexts, the same set of words may refer to quite different things. For example, 'north component of velocity' might be very difference inside an ADCP sensor than it is in a model. And then things like location and units come into play -- what does it mean to say two terms are the same? (You can look at examples like the EPIC key file definitions, whose provenance can be back-derived by those who know the domain.)

John proposed a several-part answer, based on his own experience so far and the direction we seem to be going (but maybe he just assumed that):

1. The two key attributes for device measurement parameters are the domain/substance (water, air, rocks) and the measurable property of that substance (temperature, conductivity, velocity). (See MMI guidance on this topic.)
2. Other attributes, like location, time, units, measurement method, and so on, are independent of these two primary attributes, and are not connected in a critical way to the instrument's capabilities.
3. This division of attributes does not provide the complete answer for most users' specific questions, but provide one orthogonal component. Other elements like location, time, and measurement method, can be provided by other parts of our ontology, or other parts of the data system that's managing the measurements.

Does this answer the question sufficiently?

We may not have an answer yet as to which qualifiers are and are not relevant as to data usability, but it is a necessary start.

B. Review of Vocabulary Lists

From Roy's email: "Just to report I have been unable to any work on the P012 vocabulary since last time, but it's very much still on my list."

Also clarifying another point in Roy's email, John's idea of the future for this work is that we will develop quite a number of useful parameter names as the result of this exercise. Most will probably end up in P012, some may not. But many if not all will be useful for the CF vocabulary to have; and having them in CF will make the terms more useful for this project. But the Device Ontology work will have to depend on P012 for most, if not all, of its vocabulary terms for quite some time, as terms only make their way slowly into CF.

C. Results from COMPASS Meeting

Roy also reports "I attended the COMPASS ontology workshop and found it an extremely interesting experience and highly relevant to what we are doing in this group. The start of a domain ontology incorporating platforms, instruments and parameters should be the result. We should keep a careful eye on this to make sure that our efforts are complimentary and we don't duplicate work."

To which John adds that they have collected some initial information that can be used to build an ontology for devices, and presumably have their ontology experts distilling wheat from chaff as we speak. We look forward to hearing more.

D. Discussion of GCMD Device Vocabulary Work

John sent his query to the GCMD team and will now await their reply.

This led to a discussion about the strategy for making vocabularies available for use by this team. Although MMI has GCMD science keywords available as an ontology, the document is somewhat dated.

John and Luis will look into updating the GCMD and CF ontologies served on MMI.

In the bigger picture, it would be nice to have a nice list of terms from the various controlled vocabularies that we can use for this project. This will help people understand what's available and how new terms should look. To do this we'll have to refine the existing services at the MMI site (though Roy's vocabulary will already be a nicely formatted and accessible list, so the terms in it will be nicely presentable).

John and Luis will go off and seek a mechanism to present a nice list of device parameter terms.

E. Status of Action Items

Almost all the action items relate to the items above. Surya is pursuing getting a user-focused ontology creation process to the group.

This and That

No corrections to minutes.

Closing Logistics

Next Telecon Time

The next telecon is 2 weeks from today, 2008.03.04, at 1600 GMT.

Attendees:

• Bruce Andrews
• Luis Bermudez
• Surya Durbha
• John Graybeal
• Jesper Zedlitz