Main Navigation

User login

Devices Ontology Working Group Meeting 2008.02.05

Agenda for the 2008.02.05 (1600 GMT) meeting of the Devices Ontology working group. (Minutes follow the agenda.)

A. This and That

Note Taker.
Comments/corrections of last minutes.
Comments/corrections on Agenda.
Roll Call.

B. Presentation: Dr. Femke Reitsma, Edinburgh University, COMPASS

COMPASS is an organization working on an ontologically driven knowledge infrastructure for the marine domain. One of the areas they will be considering imminently is instrument ontologies. They will talk to us about their project and their upcoming workshop (next week!), which will attempt to work toward a useful device ontology. They will also hear us talk about our project, and we may jointly get an idea of possible synergies.

C. Discussion of JCOMM/IODE Metadata Standards Meeting

The First Session of the IODE/JCOMM Forum on Oceanographic Data Management and Exchange Standards was held between 21 and 25 January 2008 at the IOC Project Office for IODE in Oostende, Belgium. (You can view an overall summary.) Several outcomes relate to our activity:

In preparation, MMI compiled a comprehensive list of device vocabularies.
John Graybeal presented an overview of 'candidate standard' instrument vocabularies
Roy Lowry of BODC/SeaDataNet presented a more comprehensive set of instrument vocabularies.
Jianping Mao of GCMD presented a description of current and pending GCMD instrument vocabulary deliverables (see next item).

These activities all potentially feed into the device ontology work, and vice-versa.

D. Discussion of GCMD Device Vocabulary Work

I hope Jianping Mao of GCMD can join us for this discussion; in any event I want to review the current status of the GCMD work, and discuss how we might best leverage it (and other possible synergistic outcomes).

E. Status of Action Items

I have updated many of the TRAC action items; if there is time, we will review pending Action Items.

See the TRAC page for the latest list of action items (due by current telecon) and list of action items from most recent telecon.

This and That

No corrections to minutes.

Femke Reitsma, COMPASS

Objective of this project is to develop a prototype in the Knowledge Infrastructure, by using ontologies as basis for discovery of data sets. They are not looking to develop the best or most comprehensive ontology, but to create a subset to demonstrate capability. Instrument ontologies is a gap in existing knowledge sets, and can be used to link data sets back to instrument. The group is looking for buy-in from communities like ours, both to help push this forward, and to get feedback on their work.

The initial stage of the work is about developing a domain ontology, which will be used to annotate science resources. They are involving a small group of domain experts, to minimize disagreement. They are looking for a small and sensible ontology, not a perfect one. The workshop next week will solicit knowledge about a small set of instruments from domain experts (while trying to leave the exact process for doing so a little bit open at this stage; they hope to make it fresh for the participants in the workshop). After generating a rough sketch of the ontology during the workshop, they will leave the workshop to create a more detailed domain ontology using that knowledge, then interact further with domain experts to refine that product.

Their method for eliciting information is based on one used in SWING project, Semantic Web Services Interoperability for Geospatial Decision Making (http://www.swing-project.org/). The use cases they dealt with are about constructions of maps, for example from OGC-service concepts to the domain ontology concepts (like the annotation task) on the fly. They had to build several domain ontologies on the fly, extracting from on-line resources. They did not use OWL, but instead used WSML (Web Services Modeling Language).

They expect to be compressing knowledge acquisition into a much smaller time for this project.

Will the organizational criteria of the ontology emerge organically or be guided? We expect it to emerge organically, driven by domain experts. The results of this will be interesting, given past 'false starts' on the MMI front.

It would be great to have out of this telecon a plan for integrating these different activities. Any ontologies produced by COMPASS would be great to feed back in to the Device Ontology group, and vice versa. One way or another, the Device Ontology group should be able to find ways to exchange information, results, and progress.

Discussion of IODE/JCOMM meeting, and SeaDataNet and GCMD activities

The meeting was very informative, and led to identification of much relevant information (see links in Agenda above). Our sense was that the participants appreciated the value that ontologies could bring to the tasks at hand, and that the typical solution would not be a single vocabulary, but would likely involve mappings between existing vocabularies. The prospects for more authoritative mappings, including in the marine sensors domain, were greatly increased as a result of the information gathered here.

There may be additional sensor vocabularies in the European community that can be identified. Roy Lowry will have a conversation with some key players, to see what they can identify.

The GCMD information pointed to by the sensor vocabularies page is not entirely relevant, and certainly not current. It is not clear why our services are not picking up the most current GCMD vocabulary, unless the GCMD services are pointing to the old content. John will pursue this with Jianping Mao; in the meantime, a note was added to the GCMD vocabularies page in MMI. Note that the GCMD vocabulary served by MMI (at marinemetadata.org/gcmd) is in OWL format, but does not represent sophisticated relationships (because these can not be trivially deduced from the provided vocabulary), and includes whole hierarchies in variable names (because otherwise variable names would be duplicated).

Roy described the SeaDatanet vocabularies of sensors. Developing these required ensuring backward compatibility with GF3, a difficult challenge. The set of 5 lists covering samples, in-situ measurements, sample processors, remote instruments, sample analyzers. These lists are governed in the SeaDataNet Task Team (someday that might change), but the list itself is public. It would be great to map the GCMD sensors to the SeaDataNet remote sensors vocabulary. Some judgment apparent in this vocabulary comes from the predecessor materials (e.g., IOC Manuals and Guides 17 Volume 2 (1987)). Note there are a lot of recently developed capabilities available through the SeaDataNet Vocabulary Server (see the links at the page linked above to access the vocabularies).

There is also a lot of new work in GCMD that folks may not be aware of; these potential capabilities surfaced in Jianping's presentations at the IODE/JCOMM meeting. The limited availability/visibility of the GCMD vocabulary and services was identified as a concern by several participants. We hope to be able to work with Jianping Mao to identify a means to collaborate and access these resources more effectively.

Status and Direction

Many of the TRAC action items have been updated, and most of the pending ones speak to the issues of having good variable names, and good instructions on how to document the variable names for each instrument. Roy and John are working on both these action items, and John hopes to publish a new set of instructions before the next telecon.

That said, are we on the right track? John is not hearing no, but not hearing yes, either. One participant offered that having something more comprehensible for instructions seems to be headed in the right direction. Another observes that a wide scope for the domain makes it difficult, if not impossible, to develop a meaningful ontology; we must narrow the scope considerably to be effective. Is the existing initial scope of instruments and variables (use case 5) narrow enough? Not clear.