[obs] Species occurrences in OBOE

[Hilmar Lapp]

Hi Matt,

thanks for this, very useful, and I wish I had had time to look more  
closely at this earlier. Some comments, specifically about the example  
for the bird observation:

1 - It seems weird to have to represent every property of a particular  
(individual in OWL) as a measurement, even if for many of those we  
know that nobody measured them, and even if someone did that there  
aren't alternative outcomes possible. Is this a modeling pattern for  
OBOE?

To they seem like an artifact of the entirely flat and redundant  
structure of DwC, and I'm not sure it'd be good practice to keep those  
artifacts in a more expressive (and semantically better defined)  
representation such as OBOE.

For example, the countryCode of Argentina is a property of the  
particular Argentina, and it is set. Also, that the province Neuquén  
is located within Argentina can perhaps be an observation one can  
make, but that knowledge should be encoded in a Gazetteer (and it is  
indeed [1]), or if that's not available, shouldn't there be a direct  
assertion between the two?  Other examples include: geodeticDatum (a  
property of the lat/long measurement), URI IDs for spatialLocation and  
samplingEvent and occurrenceID (are assignments, not measurements),  
institutionCode and collectionCode (also assignments, not  
measurements), basisOfRecord (standard or protocol, not measurement).

2 - Measurements resulting in multiple values become multiple  
measurements (presumably as a result of hasValue being a functional  
property?). Example: lat/long - this is probably from a single GPS  
reading, not two.

3 - Similar to the above, genus, class and specific epithet are not  
independent "measurements" from scientificName. One determines the  
other.

4 - Where can I find oboe-rel definitions? oboe-rel:within seems to be  
used with very heterogenous semantics (geographically located in, at  
the time of, result of).

Also, if OBOE is indeed confined to direct observational measurements  
(as Shawn recently suggested), then  those properties and assignment  
should be represented differently indeed, shouldn't they?

	-hilmar

[1] http://bioportal.bioontology.org/visualize/40651/?conceptid=GAZ%3A00004077

On Nov 24, 2010, at 6:27 PM, Matt Jones wrote:

> Hi Hilmar,
>
> I spent a bit of time with Shawn and Ben working up two Darwin Core  
> examples in OBOE.  One example is a Darwin Core record of a specimen  
> record, and the other is a Darwin Core record of a bird observation  
> where they counted birds in a survey. They both should clarify how  
> OBOE makes relationships clear and, in particular, makes the  
> contextual relationships among different observations explicit (that  
> were otherwise implicit in the Darwin Core record).
>
> The examples show how OBOE can flexibly incorporate terms from other  
> ontologies that represent, for example, Entities and  
> Characteristics.  Note that I needed ontology classes in several  
> areas, so I was lazy and just used a hypothetical 'foobar' ontology  
> namespace for classes I needed.  But the idea is that any proper OWL  
> ontology that has the right mappings to OBOEs classes can be used.
>
> https://sonet.ecoinformatics.org/observational-data-use-cases/oboe-representation-examples
>
> This approach would be relevant to use case 1 on kelp/nitrogen and  
> on use case DC2 on mammal observations.
>
> Matt
>
> On Mon, Nov 22, 2010 at 5:25 PM, Matt Jones <jones at nceas.ucsb.edu>  
> wrote:
> Hi Hilmar,
>
> Its definitely possible to represent a DarwinCore (DwC) record or  
> set of records using OBOE.  The model is general enough to  
> accomodate these types of scientific observations, including  
> specimen occurrence records, which are really just a specialized  
> form of population survey in which the specimens are at times  
> preserved.  The crux, of course, is having an appropriate OBOE  
> extension ontology that captures the essence of Darwin Core's model  
> in OWL.  We have not worked on that, and given the discussion of DwC  
> concepts that has been occurring on tdwg-content over the last few  
> weeks about the meaning of DwC terms, that might be a challenge.
>
> Nevertheless, I'll try to work up a DwC record represented in OBOE  
> as a point for discussion, and I'll send it here. And I'll probably  
> add it to the use case pages on the SONet site somehow.
>
> Matt
>
>
> On Sat, Nov 20, 2010 at 2:04 PM, Hilmar Lapp <hlapp at nescent.org>  
> wrote:
> Specimen records and species occurrence records are probably most  
> typically transported in Darwin Core (DwC). Bob Morris and I are  
> wondering whether anyone has thought yet about wrapping such a DwC  
> record in an OBOE document.
>
> The OBOE paper doesn't provide a treatment of this, and my  
> recollection from the TDWG Task Group meeting is that it hasn't been  
> thought through yet. Also, looking at OBOE, DwC terms might be used  
> to describe the Entity, but a direct wrapping does not seem obvious  
> if not impossible. That's not to say that a DwC occurrence (an  
> observation) couldn't be mapped to an equivalent OBOE document, but  
> in my recollection such a mapping has not been defined yet.
>
> Can those who are more familiar with OBOE confirm this or provide  
> pointers to where this question is being dealt with?
>
>        -hilmar
> -- 
> ===========================================================
> : Hilmar Lapp  -:- Durham, NC -:- informatics.nescent.org :
> ===========================================================
>
>
>
> _______________________________________________
> obs mailing list
> obs at ecoinformatics.org
> http://lists.nceas.ucsb.edu/ecoinformatics/mailman/listinfo/obs
>
>

-- 
===========================================================
: Hilmar Lapp  -:- Durham, NC -:- informatics.nescent.org :
===========================================================



-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.nceas.ucsb.edu/ecoinformatics/pipermail/obs/attachments/20101215/9f84a32f/attachment.html>