[seek-kr-sms] OBOE discussion: current version

Ferdinando Villa ferdinando.villa at uvm.edu
Thu Jun 22 09:10:08 PDT 2006


Hey Deana,

	I don't think there's a contradiction here. It's hard to dispute
that while observations are events (-> perdurants), observation RECORDS are
endurants, and that's precisely 1) why we collect them, and 2) what OBOE is
concerned about. If you accept my postulate that the contextualization
relationship is the translation of the where/when/how of the event into a
formal assertion (linked instances of OBOE's Observation) that allows its
representation as a dataset, your argument translates into the issue of
merging observations on the basis of the respective contexts, which is
obviously very relevant, but can be fully described within the current
framework. Ideally the "merge actor" would merge annotations and create the
annotation for the higher-level dataset, or flag incompatibilities. I think
the discussion is in part biased by the mental image of a dataset as an
observation, while we should think of an observation in its own right - a
dataset can contain multiple separate observation graphs (observations
linked by contextualization) or a single observation can be constructed from
multiple datasets. In my view, this is a representational issue, which has
its own semantics (like other things that appeared here, e.g. the
storage/number type) and that's why I proposed a separate representation
ontology. OBOE as such is concerned with the abstract semantics of the
observation. Am I missing anything?

On another note, I must make it clear that as interested I am in this
discussion, I also have to let many issues drop, as I need to keep everyone
funded here and I'm hard pressed with proposal deadlines and two guests
scientists coming to work with me in July. The dates chosen for the KR
meeting - which would be the way for me to give this discussion the
contiguous attention it deserves -  won't make me able to attend, and I
simply don't have the resources to participate fully to this discussion by
email. So I apologize in advance if I ignore important issues or do not
clarify/correct ideas that are attributed to me - just know that somewhere
in Vermont, ten fingers are almost constantly itching....

ferdinando

--
Ferdinando Villa, Associate Research Professor, Ecoinformatics
Ecoinformatics Collaboratory, Gund Inst. for Ecol. Economics and Dept. of
Botany
University of Vermont           http://ecoinformatics.uvm.edu 

> -----Original Message-----
> From: Deana Pennington [mailto:dpennington at lternet.edu] 
> Sent: Thursday, June 22, 2006 11:43 AM
> To: Ferdinando Villa
> Cc: 'Joshua Madin'; 'Serguei Krivov'; seek-kr-sms at ecoinformatics.org
> Subject: Re: [seek-kr-sms] OBOE discussion: current version
> 
> I hate to complicate this further, but I think we have to 
> deal with perdurants.  Use case scenario:  we want to combine 
> biomass data from two different field sites.  Site 1 has 
> measured biomass 2x per year (spring and fall) for 5 years, 
> and has stored all of this in a single dataset.  Site 2 has 
> measured biomass 1x per year (fall only) for 10 years that 
> partially overlap the other study, and has each year in a 
> separate dataset.  We have the data from site 1, we want to 
> query to discover any overlapping data, and then construct a 
> new integrated dataset (fall data for the overlapping years). 
>  To capture these relationships we have to recognize separate 
> events within a given dataset, which gets back to Sergey's 
> comment about relationships between rows.  We also would have 
> to annotate temporal relationships BETWEEN datasets (ick - I 
> hate to even bring this up!), unless we think the system is 
> going to be smart enough to figure out those relationships 
> from the metadata (I don't think so).  Maybe we need to keep 
> OBOE as an ontology of endurant's, but have a perdurant 
> ontology that can be used 
> to describe event relationships?   At that point, I think you 
> are trying 
> to describe how a particular set of concrete observations 
> fits into an abstract observing framework that is not 
> instantiated anywhere except in the mind of the observer, but 
> needs to be constructed by the reasoner.  
> Does this make sense to anyone???  There is an analogous 
> problem with space.  I'm thinking about the spatial problem, 
> especially with Bob Morris raising the imagery issues.  Might 
> need to get a subgroup together to start work on that.  We 
> need to consider how (if?) other data representations such as 
> imagery work under the OBOE framework.
> 
> Maybe what we need is a higher level framework that describes 
> endurants 
> and perdurants, linking space, time, and observation?   Are 
> there other 
> high level concepts that belong at that level?
> 
> Deana
> 
> 
> Ferdinando Villa wrote:
> > To express the points below (Measurement as event) more fully: the 
> > whole contextualization business is the part of the 
> Observation that 
> > relates to the event of observing. In order to describe the 
> > Observable, we must observe it, and in order to observe it, we must 
> > define a space/time/method/treatment/etc context - be 
> there, at that 
> > time, use that method etc. The Observation as a "result of 
> the event 
> > of observing" needs contextual observations in order to 
> characterize 
> > the event. So if we want to see observation as an event, it 
> is a whole 
> > new ontology we're talking about - one of perdurants.
> > Our datasets are endurants and require that OBOE is an ontology of 
> > endurants, which crystallizes the RESULT of the observation process.
> > Our hasContext relationship translates the perdurant 
> > space/time/method/etc context of the event Observation into 
> endurant 
> > Observations that can be recorded in a dataset.
> >  
> > I would also postulate that an annotation must contain one and only 
> > one Observation that is not the target of a hasContext relationship.
> > If more than one appear during an annotation, we have more than one 
> > separate datasets in the same physical storage.
> >  
> >  I totally agree and there has been some discussion about this.  I 
> > really like your use of the word "event", I think this 
> makes what we 
> > are actually doing much clearer.  For example, we Observe a 
> tree, each 
> > Measurement is an event relating to that Observation, and a 
> > Measurement is of some characteristic.  I second the move 
> changing the 
> > property between Observation and Meaurement to something like 
> > "hasMeasurementEvent".
> >
> >          
> >          
> >         Our definition, if I remember correctly, was 
> :Observation is a
> >         statement that an Observable has been observed. I think more
> >         than this is going to color OBOE with restrictions 
> it does not
> >         need to have. By the way, we model the result of the
> >         observation, not the process of the observation, and the
> >         result is not an event. To annotate a dataset we 
> don't need to
> >         know anything about the measurement except its results.  
> >          
> >
> > 
> ----------------------------------------------------------------------
> > --
> >
> > _______________________________________________
> > Seek-kr-sms mailing list
> > Seek-kr-sms at ecoinformatics.org
> > 
> http://mercury.nceas.ucsb.edu/ecoinformatics/mailman/listinfo/seek-kr-
> > sms
> >   
> 
> 
> --
> ********
> 
> Deana D. Pennington, PhD
> Long-term Ecological Research Network Office
> 
> UNM Biology Department
> MSC03  2020
> 1 University of New Mexico
> Albuquerque, NM  87131-0001
> 
> 505-277-2595 (office)
> 505-249-2604 (cell)
> 505 277-2541 (fax)
> 



More information about the Seek-kr-sms mailing list