response from the Open Annotation group

112 views
Skip to first unread message

Steve Baskauf

unread,
Feb 25, 2013, 8:38:20 PM2/25/13
to TDWG-RDF TG, J.Ke...@napier.ac.uk, Rob Guralnick
I'm copying Jessie Kennedy on this email since I pose a historical TCS question later on.
[topic background summary page: http://code.google.com/p/tdwg-rdf/wiki/OpenAnnotationAndTDWG ]

Here is a response that I got from Rob Sanderson, one of the editors of the OA Data Model.  It appears that using OA terms as I did in my example does not go beyond the scope of what is intended for the OA model.  So this seems to be an appropriate mechanism for expressing dwc:ResourceRelationship instances in RDF without the necessity of creating new DwC terms to handle the job.  What would be needed is a controlled, URI-identified vocabulary which defines the various oa:Motivation instances that would be analogous to the text values of dwc:relationshipOfResource.  Bob has provided a possible template for the SKOS and as Éamonn noted, the VoMaG may be able to facilitate the documentation of it through the system being tested on the GBIF wiki.

I think that this also means that if one asserted that

tc:TaxonRelationshipTerm rdfs:subClassOf oa:Motivation

one could use the various terms from the TDWG TaxonConcept ontology such as tc:IsCongruentTo, tc:IsChildTaxonOf, etc. as  objects of oa:motivatedBy for taxon concept relationships describe using OA.  I'm still not clear why it is better to use this annotation-style structure rather than just asserting a single triple relating the two taxa, e.g. why is it better to say:

<relationship1> a oa:Annotation;
     oa:hasBody <taxon1>;
     oa:hasTarget <taxon2>;
     oa:motivatedBy tc:IsCongruentTo.
    
rather than just saying:

<taxon1> ex:isCongruentTo <taxon2>.

where ex:isCongruentTo is an object property that I just made up but which could be part of a TDWG vocabulary successor to the TDWG TaxonConcept ontology.  Is there some historical reason in the taxon concept literature why relationships are modeled "annotation-style" in the TDWG TaxonConcept ontology?  Is it somehow a holdover from XML and the TCS standard?  I don't see the benefit of multiple triples over one triple, other than possibly allowing for provenance-tracking of the relationship assertions. 

Steve

-------- Original Message --------
Subject: Re: body which is a non-information resource
Date: Mon, 25 Feb 2013 16:33:59 -0700
From: Robert Sanderson <azar...@gmail.com>
To: Steve Baskauf <steve....@vanderbilt.edu>
CC: <public-ope...@w3.org>
References: <512B8261...@vanderbilt.edu>

Hi Steve,


On Mon, Feb 25, 2013 at 8:25 AM, Steve Baskauf
<steve....@vanderbilt.edu> wrote:

> <http://guid.mvz.org/relations/23423> a dwc:ResourceRelationship,
> oa:Annotation;
>                                      oa:hasBody
> <http://arctos.database.museum/guid/MVZ:Mamm:14523>;
>                                      oa:hasTarget
> <http://arctos.database.museum/guid/MVZ:Mamm:14524>;
>                                      oa:motivatedBy
> <http://rs.tdwg.org/relations/offpringOf>;
>                                      oa:annotatedAt "2001-09-14";
>                                      oa:annotatedBy
> <http://guid.mvz.org/agents/James_L_Patton>.
>
> <http://arctos.database.museum/guid/MVZ:Mamm:14523> a
> dcmitype:PhysicalObject, dwctype:PreservedSpecimen.
> <http://arctos.database.museum/guid/MVZ:Mamm:14524> a
> dcmitype:PhysicalObject, dwctype:PreservedSpecimen.

Very nice! :)



> There are several issues that come to my mind with this kind of use:
>
> 1.  According to the OA model description, typically the Body "is the
> comment or other descriptive resource" and the Target is a thing that "the
> Body is somehow 'about' ".

Yes, we were careful to put in weasel words like "typically",
"generally", "frequently" and so on when trying to define what we
meant by Annotation, Body and Target.  The sort of thing that most
people think of when they hear "annotation" is a comment about
something, and we didn't want to define it such that it would be
surprising.

However we definitely recognize other use cases, such as linking,
classifying and identifying where the body is in no way "about" the
target.
What we want to avoid, however, is re-inventing RDF inside RDF where
an Annotation ends up just being a triple with subject (body),
predicate (motivation) and object (target).

> In my example the Body is a dead mammal which is
> a non-information resource and is in no way descriptive.  The Body is
> related to, but not really "about" the Target, although one of the listed
> instances of oa:Motivation is oa:linking, and asserting that one organism is
> the offspring of another is a sort of linking.

Yes, that's perfectly okay.


> 2. I'm a little confused about what one is doing when one creates an
> Annotation.  I think that creating an annotation is the act of asserting
> that there is a connection between the two resources.  However, the
> description of the oa:Motivation class says that a Motivation instance is
> the reason for the creation of the Annotation instance, NOT the reason for
> the creation the relationship which the Annotation instance documents.  Are
> the various items on the list of instances of oa:Motivation things that we
> are saying an annotating agent has done?  Or are those things that we are
> saying that the annotating agent is asserting has been done?  For example,
> if an agent creates an Annotation instance with motivation oa:commenting, do
> we assume that agent has actually created the comment or that the agent is
> just documenting that a comment has been created by someone else?  In my
> example above, the annotating agent cannot have a role in the creation of
> the relationship between the Body and Target.  The agent is simply recording
> the existence of the relationship.

That's a very good point, and it isn't clear in the documentation, I
agree.  The Motivation is information about the relationship between
the resources, and not necessarily to do with the agent that created
the Annotation.  For example, I could create an Annotation where
someone else's YouTube video is the body, and the target is an image
on the web.  I didn't do the commenting, but I'm asserting that the
reason the annotation was created was to model the fact that the body
resource is commenting on the target resource -- this seems to be
exactly equivalent with your description: The agent is simply
recording the existence of the relationship.


> Using the OA model to document dwc:ResourceRelationship instances is very
> appealing to me.  I'm just not sure if it is appropriate to use OA to
> describe relationships that may not fit the description of oa:Annotation
> instances, and with Body resources that are non-information resources.

I think it should be fine :)

Hope that helps!

Rob

.


-- 
Steven J. Baskauf, Ph.D., Senior Lecturer
Vanderbilt University Dept. of Biological Sciences

postal mail address:
VU Station B 351634
Nashville, TN  37235-1634,  U.S.A.

delivery address:
2125 Stevenson Center
1161 21st Ave., S.
Nashville, TN 37235

office: 2128 Stevenson Center
phone: (615) 343-4582,  fax: (615) 343-6707
http://bioimages.vanderbilt.edu

Paul J. Morris

unread,
Feb 25, 2013, 8:48:33 PM2/25/13
to tdwg...@googlegroups.com, steve....@vanderbilt.edu, J.Ke...@napier.ac.uk, Rob Guralnick
On Mon, 25 Feb 2013 19:38:20 -0600
Steve Baskauf <steve....@vanderbilt.edu> wrote:
> <relationship1> a oa:Annotation;
> oa:hasBody <taxon1>;
> oa:hasTarget <taxon2>;
> oa:motivatedBy tc:IsCongruentTo.

I don't understand what is being said in the formulation above. It doesn't seem to correspond to any way in which we've used AO or OA.

> rather than just saying:
>
> <taxon1> ex:isCongruentTo <taxon2>.

This would seem to be an appropriate formulation. It can't carry the metadata about who, when, and why this assertion was made that an annotation can carry.

If an annotation were to be used, I would expect it to say something in the form of:

oa:Annotation;
oa:hasBody
relationship1
<taxon1> ex:isCongruentTo <taxon2>;
oa:hasTarget
Some data set with some selector.
oa:motivatedBy
Annotator's desire to express a relationship.
oad:hasEvidence
The evidence supporting the congruence of taxa 1 and 2.


-Paul
--
Paul J. Morris
Biodiversity Informatics Manager
Harvard University Herbaria/Museum of Comparative Zoölogy
mo...@morris.net AA3SD PGP public key available

Steve Baskauf

unread,
Feb 25, 2013, 9:17:36 PM2/25/13
to Paul J. Morris, tdwg...@googlegroups.com, J.Ke...@napier.ac.uk, Rob Guralnick
Check out the diagrams and comparison table at http://code.google.com/p/tdwg-rdf/wiki/OpenAnnotationAndTDWG

The way that TDWG TaxonConcept ontology models relationships is analogous to the OA model, except that tc:fromTaxon is used instead of oa:hasBody and tc:toTaxon is used instead of oa:hasTarget.  Thus using the TaxonConcept ontology terms, I would say:

<relationship1> a tc:Relationship;
     tc:fromTaxon <taxon1>;
     tc:toTaxon <taxon2>;
     tc:relationshipCategory tc:IsCongruentTo.

It's a different set of terms, but to a machine it would be essentially equivalent the example below.  Based on the email I got from Rob Sanderson, the OA model is broad enough to encompass non-information resources (such as taxon concepts) and motivations of virtually any sort that would relate two resources.  So I take that to mean that the RDF I wrote below is a valid use of OA and OA is (or probably will be) more well-known than the TDWG ontology.  However, what I'm wondering is why I would want to use either the RDF above or the RDF below when I could just use a simple predicate and one triple.  Obviously Roger Hyam thought that was the best way to set it up, but he never answers my emails, so I can't ask him why he wrote the ontology that way.

Steve


Paul J. Morris wrote:
On Mon, 25 Feb 2013 19:38:20 -0600
Steve Baskauf <steve....@vanderbilt.edu> wrote:
  
<relationship1> a oa:Annotation;
     oa:hasBody <taxon1>;
     oa:hasTarget <taxon2>;
     oa:motivatedBy tc:IsCongruentTo.
    
I don't understand what is being said in the formulation above.  It doesn't seem to correspond to any way in which we've used AO or OA.

  
rather than just saying:

<taxon1> ex:isCongruentTo <taxon2>.
    
This would seem to be an appropriate formulation.  It can't carry the metadata about who, when, and why this assertion was made that an annotation can carry.

If an annotation were to be used, I would expect it to say something in the form of: 

oa:Annotation;
      oa:hasBody 
            relationship1 
               <taxon1> ex:isCongruentTo <taxon2>;
      oa:hasTarget 
            Some data set with some selector.
      oa:motivatedBy
            Annotator's desire to express a relationship.
      oad:hasEvidence
            The evidence supporting the congruence of taxa 1 and 2.


-Paul
  

-- 
Steven J. Baskauf, Ph.D., Senior Lecturer
Vanderbilt University Dept. of Biological Sciences

postal mail address:
VU Station B 351634
Nashville, TN  37235-1634,  U.S.A.

delivery address:
2125 Stevenson Center
1161 21st Ave., S.
Nashville, TN 37235

office: 2128 Stevenson Center
phone: (615) 343-4582,  fax: (615) 343-6707
http://bioimages.vanderbilt.edu

Hilmar Lapp

unread,
Feb 25, 2013, 9:47:15 PM2/25/13
to tdwg...@googlegroups.com, J.Ke...@napier.ac.uk, Rob Guralnick

On Feb 25, 2013, at 8:38 PM, Steve Baskauf wrote:

<relationship1> a oa:Annotation;
     oa:hasBody <taxon1>;
     oa:hasTarget <taxon2>;
     oa:motivatedBy tc:IsCongruentTo.
     
rather than just saying:

<taxon1> ex:isCongruentTo <taxon2>.

There's a key piece in Rob Sanderson's response, which until I see a good argument to the contrary my common sense tells applies here: don't reinvent RDF within RDF. I.e., don't use oa:Annotation to say something that could just as well be said in a triple directly.

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



Steve Baskauf

unread,
Feb 25, 2013, 10:36:34 PM2/25/13
to tdwg...@googlegroups.com, J.Ke...@napier.ac.uk, Rob Guralnick
Hilmar,
The reason why I'm bringing this up here is because a major task that the VoMaG Task Group has taken on [1] is to come up with some recommendations on what to do with the TDWG Ontology and I've gotten drafted to facilitate the discussion on that topic.  Based on my informal observations, there are only two major parts of the TDWG ontology that anybody seems to have ever actually used for anything: the NCD ontology and the TaxonConcept/TaxonName ontologies.  What I'm saying here is that the TaxonConcept ontology [2] essentially does what you are saying common sense dictates against: reinventing RDF within RDF by using the tc:toTaxon and tc:fromTaxon terms.  I'm with you, I don't see any sense in that.  But apparently Roger Hyam and possibly other people did some sense in it and I'd be curious to know why.  If this really isn't a good approach, then is our (the TDWG RDF Task Group) advice to the VoMaG group that most of the TDWG TaxonConcept ontology be depricated? 

The only possible reason to use OA as a substitute for the TaxonConcept ontology terms are that OA may be more well-known.  If the "tc:toTaxon" and "tc:fromTaxon" approach is a waste of time, it's not going to be better to substitute oa:hasBody and oa:hasTarget.

Steve

[1] http://community.gbif.org/pg/pages/view/29079/section-1-a-review-of-the-tdwg-ontologies
[2] http://code.google.com/p/tdwg-ontology/source/browse/trunk/ontology/voc/TaxonConcept.rdf

Hilmar Lapp wrote:
--
You received this message because you are subscribed to the Google Groups "TDWG RDF/OWL Task Group" group.
To unsubscribe from this group and stop receiving emails from it, send an email to tdwg-rdf+u...@googlegroups.com.
For more options, visit https://groups.google.com/groups/opt_out.
 
 

Nico Franz

unread,
Feb 25, 2013, 11:02:18 PM2/25/13
to tdwg...@googlegroups.com, J.Ke...@napier.ac.uk, Rob Guralnick
Hi Steve:

   Jessie will know better; I personally don't recall the motivation part (not that it is "wrong"; I just don't recall it being there at the time). What was there was an AccordingTo to accommodate for the reality of these relationships being done by people, where disagreements (multiple person-specific assessments about the same pair of concepts) are possible. Also, in this particular context, I would understand "motivation" to point to reasons for establishing a relationship (either why it needs to be established, or why it has a particular relationship). "IsCongruentTo" is an RCC relation (http://en.wikipedia.org/wiki/Region_connection_calculus); it is not a "motivation" in either sense (why needed?, why that?) that would let another actor lead to comprehend the nature/underlying grounds for a particular relationship assertion.

   So that could mean a few minor modifications (if the representing the original TCS is the goal).

Best,

Nico


Nico M. Franz, Ph.D.
Associate Professor & Curator of Insects
School of Life Sciences
PO Box 874501
Arizona State University
Tempe, AZ 85287-4501

Office: (480) 965-2036
Collection: (480) 965-2850
Fax: (480) 965-6899
E-mail: nico.franz @ asu.edu

Franz Lab: http://franz.lab.asu.edu/
ASUHIC: http://symbiota1.acis.ufl.edu/scan/portal/collections/misc/collprofiles.php?collid=1
WoNA: http://symbiota1.acis.ufl.edu/scan/portal/checklists/checklist.php?cl=1

  

Bob Morris

unread,
Feb 25, 2013, 11:29:36 PM2/25/13
to tdwg...@googlegroups.com, J.Ke...@napier.ac.uk, Rob Guralnick
(Perhaps a red-herring): As Paul remarked, in this case using OA
provides for provenance of the assertion of the relation, as well as
the serialization with some specified predicates, and with a defined
mapping to W3C PROV http://www.w3.org/TR/prov-overview/ . To whatever
extent VoMaG cares about such provenance and standardizing it for the
TDWG community, it would need \some/ standard predicates to address
provenance, and maybe needs something as heavy as PROV anyway.
Robert A. Morris

Emeritus Professor of Computer Science
UMASS-Boston
100 Morrissey Blvd
Boston, MA 02125-3390

IT Staff
Filtered Push Project
Harvard University Herbaria
Harvard University

email: morri...@gmail.com
web: http://efg.cs.umb.edu/
web: http://wiki.filteredpush.org
http://www.cs.umb.edu/~ram
===
The content of this communication is made entirely on my
own behalf and in no way should be deemed to express
official positions of The University of Massachusetts at Boston or
Harvard University.

Steve Baskauf

unread,
Feb 25, 2013, 11:38:09 PM2/25/13
to tdwg...@googlegroups.com
Nico,
The "motivation" part is the term that the Open Annotation model uses to describe the relationship between the two resources that are being "annotated".  I think they call the term "motivation" because the kind of relationships that they have in mind are annotations that primarily document resources that were created to respond to some existing resources (so the "motivation" is the reason why they were created - to comment, to edit, to revise, etc.).  But if I'm understanding Rob Sanderson correctly, an oa:Motivation instance can be any relationship between resources, not just relationships that were created by some human motivation. 

In the TaxonConcept ontology, the class whose instance describe the relationship between two resources is tc:TaxonRelationshipTerm .  It's analogous to oa:Motivation in OA and I suppose that tc:TaxonRelationshipTerm might be considered a subclass of oa:Motivation, although I'm no longer too sure about anything here! :-)

Steve

Steve Baskauf

unread,
Feb 25, 2013, 11:57:24 PM2/25/13
to tdwg...@googlegroups.com, J.Ke...@napier.ac.uk, Rob Guralnick
Well, I think you may have hit the nail on the head here: "to the extent that VoMaG cares about provenance", or maybe we should say to the extent that TDWG cares about provenance.  With respect to the TDWG TaxonConcept ontology, the extent to which it matters probably depends on how TDWG intends for TaxonConcepts to be managed.  This brings to mind the previous discussion about the TaxonConcept ontology that Rich Pyle contributed to a few months ago.  If there will be many players in the taxon concept game, each asserting relationships among various taxon concepts, then tracking the provenance of those assertions would probably be very important to aggregators who collect taxon concept assertions from many sources.  On the other hand if one big organization (like the vaporware GNUB) manages all of the TaxonConcepts/TNUs for the planet, then exposing the provenance as RDF isn't really that important.  "Where did the assertion come from? GNUB, of course, they all do."  GNUB's internal database can track the sources and dates without exposing them all as RDF. 

I think that TCS and the TaxonConcept ontology were written in an era where perhaps it was assumed that many providers would be generating taxon concept relationships and therefore an RDF structure that was conducive to tracking provenance was adopted.  A simpler model where taxon concepts are related by a single property in a single triple might be more appropriate in a "one-stop-shopping" world where taxon concept relationships are provided at one source.  That is, if one-stop-shopping for TaxonConcepts/TNUs ever comes into being.

Steve

Roderic Page

unread,
Feb 26, 2013, 2:01:26 AM2/26/13
to tdwg...@googlegroups.com, J.Ke...@napier.ac.uk, Rob Guralnick
I can't speak for Roger Hyam, but I suspect the reason is as you suggest:

> I don't see the benefit of multiple triples over one triple, other than possibly allowing for provenance-tracking of the relationship assertions.

The TDWG taxon concept vocabulary covers a multitude of possible relationships. If we think simply in terms of a parent-child hierarchy, then a simple triple makes sense. For example, Uniprot has this triple for the shrimp Rimicaris exoculata:

http://purl.uniprot.org/taxonomy/71621 rdfs:subClassOf http://purl.uniprot.org/taxonomy/71620 (Uniprot doesn't use the TDWG vocabulary)

Simples. I guess in this context we are happy to assume that the NCBI has put Rimicaris exoculata in the genus Rimicaris (http://purl.uniprot.org/taxonomy/71620) for a good reason.

But what about an assertion that Rimicaris aurantiaca is a synonym of Rimicaris exoculata? How do I make that assertion and link it to evidence? I suspect Roger wanted to enable this, but at the same time avoid reification (in other words, so that the assertion of synonymy can have it's own URI). So, presumably I can say something like this (in bastard RDF):

<TaxonRelationshipTerm rdf:resource="http://xxx">
fromTaxon: <Rimicaris aurantiaca>
toTaxon: <Rimicaris exoculata>
TaxonConcept:relationshipCategory <IsSynonymFor>
</TaxonRelationshipTerm>

Given that this has a URI ("http://xxx") I can now add statements about the evidence (in this case, published in http://www.ncbi.nlm.nih.gov/pubmed/9628005 ) for the assertion that Rimicaris aurantiaca is a synonym of Rimicaris exoculata.

Regards

Rod


Éamonn Ó Tuama [GBIF]

unread,
Feb 26, 2013, 4:47:52 AM2/26/13
to tdwg...@googlegroups.com, J.Ke...@napier.ac.uk, Rob Guralnick

Can someone knowledgeable comment further on why RDF reification is not adopted in preference to creating special schemes for annotation (OA)/provenance (PROV). I note on Wikipedia:

“The power of the reification vocabulary in RDF is restricted by the lack of a built-in means for assigning URIrefs to statements, so in order to express "provenance" information of this kind in RDF, one has to use some mechanism (outside of RDF) to assign URIs to individual RDF statements”. Does not the same apply to OA and PROV?

-Éamonn

--

Roderic Page

unread,
Feb 26, 2013, 4:55:03 AM2/26/13
to Kennedy, Jessie, tdwg...@googlegroups.com, Rob Guralnick
Hi Jessie,

>
> The relationships being discussed are not parent child relationships -
> these are handled differently in TCS - with is child of or has children
> terms.

Unless I'm mistaken the RDF vocabulary Roger created treats child relationships a kind of TaxonConcept:Relationship

e.g., this snippet

<TaxonConcept:hasRelationship>
<TaxonConcept:Relationship>
<TaxonConcept:relationshipCategory rdf:resource="http://rs.tdwg.org/ontology/voc/TaxonConcept#IsChildTaxonOf"/>
<TaxonConcept:fromTaxon rdf:resource="urn:lsid:catalogueoflife.org:taxon:f5f198dc-29c1-102b-9a4a-00304854f820:ac2008"/>
<TaxonConcept:toTaxon rdf:resource=""/>
</TaxonConcept:Relationship>
</TaxonConcept:hasRelationship>

(from a cached LSID resolution http://darwin.zoology.gla.ac.uk/~rpage/lsid/tester/cache/catalogueoflife.org/08003831cd301942e0a1e1fd181c4692.rdf )

>
>>
>> Simples. I guess in this context we are happy to assume that the NCBI has
>> put Rimicaris exoculata in the genus Rimicaris
>> (http://purl.uniprot.org/taxonomy/71620) for a good reason.
>
> Yes at the time but did they get it right and what if someone else doing a
> more thorough examination thinks differently - how does that get recorded?
> What if they change their mind - does it matter that people used different
> information in the past - I think for most analysis people are now saying
> provenance is important and we should be able to replicate our
> data/results.

I was simply giving a example of real-world RDF that I think appeals to Hilmar. I suspect for most parent-child relationships nobody will be able to say exactly why they are what they are, unless they have some sort of versioning system recording the edits. NCBI sometimes stores citations and notes on particular nodes, but usually it is "because we say so". For a lot of things that is all we need, in that classifications are tools for navigation, IMHO they aren't "real" in any meaningful sense (if you want that you need phylogenies).

>
>>
>> But what about an assertion that Rimicaris aurantiaca is a synonym of
>> Rimicaris exoculata? How do I make that assertion and link it to
>> evidence? I suspect Roger wanted to enable this, but at the same time
>> avoid reification (in other words, so that the assertion of synonymy can
>> have it's own URI). So, presumably I can say something like this (in
>> bastard RDF):
>
> What does synonym mean - is that exact enough? Might be all you have but
> is it what we should be recording?

If that's all we have, yes, we should just record that. We actually know more about this particular example (genetic evidence suggests "Rimicaris aurantiaca" are simply juvenile R. exoculata) so could be more precise (or use the Pubmed id as a proxy for the evidence i.e., "want to know more, read this").

>
>>
>> <TaxonRelationshipTerm rdf:resource="http://xxx">
>> fromTaxon: <Rimicaris aurantiaca>
>> toTaxon: <Rimicaris exoculata>
>> TaxonConcept:relationshipCategory <IsSynonymFor>
>> </TaxonRelationshipTerm>
>>
>> Given that this has a URI ("http://xxx") I can now add statements about
>> the evidence (in this case, published in
>> http://www.ncbi.nlm.nih.gov/pubmed/9628005 ) for the assertion that
>> Rimicaris aurantiaca is a synonym of Rimicaris exoculata.
>>
>> Regards
>>
>> Rod
>>
> Back to holidayŠ It's difficult to know whether to go for the simple life
> or start to build the best you canŠ..

My own preference is simply so we actually get to build something ;)

Regards

Rod



>
>>
>>
>>
>
>
> Edinburgh Napier University is one of Scotland's top universities for graduate employability. 93.6% of graduates are in work or further study within six months of leaving. The Telegraph newspaper named us as one of the "top ten UK universities for getting a job" in 2012. This university is also proud winner of the Queen's Anniversary Prize for Higher and Further Education 2009, awarded for innovative housing construction for environmental benefit and quality of life.
>
> This message is intended for the addressee(s) only
> and should not be read, copied or disclosed to anyone else outwith the University without the permission of the sender. It is your responsibility to ensure that this message and any attachments are scanned for viruses or other defects.
> Edinburgh Napier University does not accept liability for any loss or
> damage which may result from this email or any attachment, or for errors or omissions arising after it was sent. Email is not a secure medium. Email entering the University's system is subject to routine monitoring and filtering by the University.
>
> Edinburgh Napier University is a registered Scottish
> charity.
> Registration number SC018373
>
>

Roderic Page

unread,
Feb 26, 2013, 5:03:42 AM2/26/13
to tdwg...@googlegroups.com, J.Ke...@napier.ac.uk, Rob Guralnick
Based on my hazy recollection from the days I cared more about RDF than I do now, the issue was exactly as Wikipedia describes, how do you stick an identifier on a triple? My sense is that reification makes (made?) for messy SPARQL queries and has fallen out of favour (see http://answers.semanticweb.com/questions/206/recipes-for-having-reification-without-the-rdf-reification and http://answers.semanticweb.com/questions/2398/which-type-of-reification-do-you-use ).

If the statement of, say, synonymy is itself a "thing" (e.g. an instance of a class) then it gets a URI and can be the subject of other statements in the normal way.

Regards

Rod

Kennedy, Jessie

unread,
Feb 26, 2013, 4:29:29 AM2/26/13
to Steve Baskauf, tdwg...@googlegroups.com, Rob Guralnick
Hi Steve

Am on holiday at present :-) but thought I could give a quick response….sorry if it's not well constructed….

The TCS terms were indeed to allow for the provenance of who said what relationship existed between two taxa. Given that the whole concept issue came down to people having different opinions about the extent/circumscription of a taxon the case was made that different people would have different opinions about the relationships between different taxa. Indeed it was put forward (no provenance held in my head!) that some people define taxa in terms of relationships to other taxa. So someone might choose to define a new taxon as being related in particular ways to existing taxa as described by other taxonomists.

Hope this is making sense…

So there were 2 main different types of relationships – those that were described within the definition of a taxon by some taxonomist – so you could assume the relationship was by that person, and those that were described by someone trying to make sense of the relationships between existing taxa described by someone else.
Both types had the range of relationships- - congruent to etc.

If an aggregator is defining a set of "new" taxa as agreed by them and each of the taxa they described had UIDs of their own then any relationship that they specified between their own taxa could be assumed to be by them. They would then need to make congruent relationships to existing taxa to reuse the originals. However if they reused the original taxon concepts and the original UIDs then if they wanted to related them to another taxon then they would need to say it was them making the relationship.

I think if you miss out the provenance all together then in the future you'll be in a situation like we are at present not knowing if the name means what we think it meant – especially if people do have differing opinions on how taxa are related….

I don't know what was in Roger's mind but am sure it would be related to what I've just said. There wasn't an OA group at the time when this was done, so maybe we were just providing the provence we think is needed that is being seen as important in a more general sense now.

Sorry if this is not clear….all I can do and not been keeping up with the RDF stuff sorry…

Jessie


From: Steve Baskauf <steve....@vanderbilt.edu>
Organization: Vanderbilt University Dept. of Biological Sciences
Date: Mon, 25 Feb 2013 22:57:24 -0600
To: "tdwg...@googlegroups.com" <tdwg...@googlegroups.com>
Cc: Jessie Kennedy <j.ke...@napier.ac.uk>, Rob Guralnick <Robert.G...@colorado.edu>
Subject: Re: [tdwg-rdf: 146] response from the Open Annotation group

Well, I think you may have hit the nail on the head here: "to the extent that VoMaG cares about provenance", or maybe we should say to the extent that TDWG cares about provenance.  With respect to the TDWGTaxonConcept ontology, the extent to which it matters probably depends on how TDWG intends for TaxonConcepts to be managed.  This brings to mind the previous discussion about the TaxonConcept ontology that Rich Pyle contributed to a few months ago.  If there will be many players in the taxon concept game, each asserting relationships among various taxon concepts, then tracking the provenance of those assertions would probably be very important to aggregators who collect taxon concept assertions from many sources.  On the other hand if one big organization (like the vaporware GNUB) manages all of the TaxonConcepts/TNUs for the planet, then exposing the provenance as RDF isn't really that important.  "Where did the assertion come from? GNUB, of course, they all do."  GNUB's internal database can track the sources and dates without exposing them all as RDF. 

I think that TCS and the TaxonConcept ontology were written in an era where perhaps it was assumed that many providers would be generating taxon concept relationships and therefore an RDF structure that was conducive to tracking provenance was adopted.  A simpler model where taxon concepts are related by a single property in a single triple might be more appropriate in a "one-stop-shopping" world where taxon concept relationships are provided at one source.  That is, if one-stop-shopping for TaxonConcepts/TNUs ever comes into being.

Steve

Bob Morris wrote:
(Perhaps a red-herring): As Paul remarked, in this case using OA
provides for provenance of the assertion of the relation, as well as
the serialization with some specified predicates, and with a defined
mapping to W3C PROV http://www.w3.org/TR/prov-overview/ .  To whatever
extent VoMaG cares about such provenance and standardizing it for the
TDWG community, it would need \some/ standard predicates to address
provenance, and maybe needs something as heavy as PROV anyway.

On Mon, Feb 25, 2013 at 10:36 PM, Steve Baskauf
<steve....@vanderbilt.edu> wrote:
  
Hilmar,
The reason why I'm bringing this up here is because a major task that the
VoMaG Task Group has taken on [1] is to come up with some recommendations on
what to do with the TDWG Ontology and I've gotten drafted to facilitate thediscussion on that topic.  Based on my informal observations, there are only
two major parts of the TDWG ontology that anybody seems to have ever
actually used for anything: the NCD ontology and the TaxonConcept/TaxonNameontologies.  What I'm saying here is that the TaxonConcept ontology [2]
essentially does what you are saying common sense dictates against:
reinventing RDF within RDF by using the tc:toTaxon and tc:fromTaxon terms.
I'm with you, I don't see any sense in that.  But apparently Roger Hyam andpossibly other people did some sense in it and I'd be curious to know why.
If this really isn't a good approach, then is our (the TDWG RDF Task Group)advice to the VoMaG group that most of the TDWG TaxonConcept ontology be
depricated?

The only possible reason to use OA as a substitute for the TaxonConcept
ontology terms are that OA may be more well-known.  If the "tc:toTaxon" and"tc:fromTaxon" approach is a waste of time, it's not going to be better to
substitute oa:hasBody and oa:hasTarget.

Steve

[1]


Hilmar Lapp wrote:


On Feb 25, 2013, at 8:38 PM, Steve Baskauf wrote:

<relationship1> a oa:Annotation;
     oa:hasBody <taxon1>;
     oa:hasTarget <taxon2>;
     oa:motivatedBy tc:IsCongruentTo.

rather than just saying:

<taxon1> ex:isCongruentTo <taxon2>.


There's a key piece in Rob Sanderson's response, which until I see a good
argument to the contrary my common sense tells applies here: don't reinventRDF within RDF. I.e., don't use oa:Annotation to say something that could
just as well be said in a triple directly.

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



--
You received this message because you are subscribed to the Google Groups
"TDWG RDF/OWL Task Group" group.
To unsubscribe from this group and stop receiving emails from it, send an
email to tdwg-rdf+u...@googlegroups.com.
For more options, visit https://groups.google.com/groups/opt_out.




--
Steven J. Baskauf, Ph.D., Senior Lecturer
Vanderbilt University Dept. of Biological Sciences

postal mail address:
VU Station B 351634
Nashville, TN  37235-1634,  U.S.A.

delivery address:
2125 Stevenson Center
1161 21st Ave., S.
Nashville, TN 37235

office: 2128 Stevenson Center
phone: (615) 343-4582,  fax: (615) 343-6707

Kennedy, Jessie

unread,
Feb 26, 2013, 4:41:12 AM2/26/13
to Roderic Page, tdwg...@googlegroups.com, Rob Guralnick
I think Rod's response is oversimplifying what people told me the
situation was and what they wanted to be able to do - probably not what
Rod wants to do thoughŠ.


On 26/02/2013 07:01, "Roderic Page" <rdm...@gmail.com> wrote:

>I can't speak for Roger Hyam, but I suspect the reason is as you suggest:
>
>> I don't see the benefit of multiple triples over one triple, other
>>than possibly allowing for provenance-tracking of the relationship
>>assertions.
>
>The TDWG taxon concept vocabulary covers a multitude of possible
>relationships. If we think simply in terms of a parent-child hierarchy,
>then a simple triple makes sense. For example, Uniprot has this triple
>for the shrimp Rimicaris exoculata:
>
>http://purl.uniprot.org/taxonomy/71621 rdfs:subClassOf
>http://purl.uniprot.org/taxonomy/71620 (Uniprot doesn't use the TDWG
>vocabulary)

The relationships being discussed are not parent child relationships -
these are handled differently in TCS - with is child of or has children
terms.

>
>Simples. I guess in this context we are happy to assume that the NCBI has
>put Rimicaris exoculata in the genus Rimicaris
>(http://purl.uniprot.org/taxonomy/71620) for a good reason.

Yes at the time but did they get it right and what if someone else doing a
more thorough examination thinks differently - how does that get recorded?
What if they change their mind - does it matter that people used different
information in the past - I think for most analysis people are now saying
provenance is important and we should be able to replicate our
data/results.

>
>But what about an assertion that Rimicaris aurantiaca is a synonym of
>Rimicaris exoculata? How do I make that assertion and link it to
>evidence? I suspect Roger wanted to enable this, but at the same time
>avoid reification (in other words, so that the assertion of synonymy can
>have it's own URI). So, presumably I can say something like this (in
>bastard RDF):

What does synonym mean - is that exact enough? Might be all you have but
is it what we should be recording?

>
><TaxonRelationshipTerm rdf:resource="http://xxx">
> fromTaxon: <Rimicaris aurantiaca>
> toTaxon: <Rimicaris exoculata>
> TaxonConcept:relationshipCategory <IsSynonymFor>
></TaxonRelationshipTerm>
>
>Given that this has a URI ("http://xxx") I can now add statements about
>the evidence (in this case, published in
>http://www.ncbi.nlm.nih.gov/pubmed/9628005 ) for the assertion that
>Rimicaris aurantiaca is a synonym of Rimicaris exoculata.
>
>Regards
>
>Rod
>
Back to holidayŠ It's difficult to know whether to go for the simple life
or start to build the best you canŠ..

>
>
>


Roderic Page

unread,
Feb 26, 2013, 8:07:12 AM2/26/13
to tdwg...@googlegroups.com, Steve Baskauf, Rob Guralnick
My take on TaxonConcept vocabularies is that they are typically over engineered in an attempt to capture the multitude of different ways taxonomists mess with classifications.

Frankly, the only TaxonConcepts I think are worth caring about are those connected to data, and at the scale of all of life this means GBIF and NCBI. These are the only large-scale classifications that are connected to stuff (occurrences and sequences, respectively) we can compute over. We can do science with GBIF and NCBI, other sources of taxon concepts, not so much.

So, do we have a one-stop shop for concepts? Almost. We have two "shops", GBIF and NCBI. IMHO much of the taxon concept debate has been sidetracked trying to model what taxonomists do in gory detail, that way lies madness. 

Regards

Rod

greg whitbread

unread,
Feb 26, 2013, 8:42:54 AM2/26/13
to tdwg...@googlegroups.com
My turn to say your over simplifying things Rod ... or maybe missing the point.

The classification presents only one kind of taxon relationship and everything about GBIF and NCBI is an ephemeral guess.  Science requires persistence, vocabularies that support research and the ability to link back to original work and data attached.   Concepts and classifications are taxonomy's tools of trade.  Are you saying that it is all irrelevant ... already done?

greg

If you have received this transmission in error please notify us immediately by return e-mail and delete all copies. If this e-mail or any attachments have been sent to you in error, that error does not constitute waiver of any confidentiality, privilege or copyright in respect of information in the e-mail or attachments. Please consider the environment before printing this email.

--
You received this message because you are subscribed to the Google Groups "TDWG RDF/OWL Task Group" group.
To unsubscribe from this group and stop receiving emails from it, send an email to tdwg-rdf+u...@googlegroups.com.
For more options, visit https://groups.google.com/groups/opt_out.
 
 



--
Greg Whitbread
Australian National Botanic Gardens
Australian National Herbarium
+61 2 62509482
g...@anbg.gov.au

Hilmar Lapp

unread,
Feb 26, 2013, 9:00:32 AM2/26/13
to tdwg...@googlegroups.com, tdwg...@googlegroups.com, <J.Kennedy@napier.ac.uk>, Rob Guralnick
That's about right. Note also that OWL doesn't support reification (other than through modeling the assertion itself, as here). 

-hilmar

Sent with a tap.

Roderic Page

unread,
Feb 26, 2013, 9:08:24 AM2/26/13
to tdwg...@googlegroups.com
I guess I'm saying that I view classifications as simply tools for navigation, they are a way for organising data. Classifications serve the same function as, say the Dewey library classification scheme. They bear at most a passing resemblance to, say, the phylogeny for a set of taxa (or, more correctly, the cloud of gene trees that we dignify with a name). If we were serious about classifications being important we'd focus on phylogenies (and discover that most scales the notion of a "tree of life" starts to fade away). But I digress.

As for science requiring persistence, I'm all for persistence and links to things that persist (e.g., data, literature). Indeed, this is something of an obsession of mine. But classifications are as meaningful as how I arrange my files on my hard disk. Indeed, I'd argue we should treat classifications as exactly that, lets put GBIF and NCBI into github and use version control to track the changes.

I guess I don't see the goal to be modelling what taxonomist do, nor replicating their tools of trade. The landscape has changed too much for that to be a tenable approach. 

Regards

Rod

Steve Baskauf

unread,
Feb 26, 2013, 9:34:28 AM2/26/13
to tdwg...@googlegroups.com, Rob Guralnick, J.Ke...@napier.ac.uk
I think that my basic questions have been answered now:

1. Why is the TDWG TaxonConcept ontology model structured using tc:toTaxon and tc:fromTaxon rather than by simply defining object properties to express the relationships in a single triple?  Answer: to allow for tracking provenance and to allow the assertion to be linked to other resources (e.g. citations). 

2. Is there a reason to keep the TaxonConcept ontology?  Answer: I think it's yes, because it is capable of doing what it sets out to do and it is being used by at least one major provider (APNI).  Although it apparently could be replaced with the OA model, I'm not sure that I see a compelling reason to do that.  I think recasting the model with oa:hasBody and oa:hasTarget just confuses the issue.  (That's a different situation from the Darwin Core ResourceRelationship class which does not have terms that will "work" as RDF.)

I don't think that our task here is to decide if the TaxonConcept is the best or most efficient way to express relationships between taxa.  I think that decision will be made by the "market", i.e. if it works people will use it and if it doesn't people, will use something else, although at the moment I don't see what the "something else" is which can capture the kind of information that taxonomists want. 

Thanks for all of the input on this topic.  I'll try to summarize on the wiki the thread thus far, along with any subsequent comments that people may add.

Steve

Nico Franz

unread,
Feb 26, 2013, 9:53:44 AM2/26/13
to tdwg...@googlegroups.com
I find none of this provocative at all. ;o)

Nico

Roderic Page

unread,
Feb 26, 2013, 10:04:08 AM2/26/13
to tdwg...@googlegroups.com
On 26 Feb 2013, at 14:53, Nico Franz wrote:

I find none of this provocative at all. ;o)

Then I'm clearly not trying hard enough ;)

Rod


Hilmar Lapp

unread,
Feb 26, 2013, 10:21:57 AM2/26/13
to tdwg...@googlegroups.com, Steve Baskauf, Rob Guralnick
Maybe I'm abusing the TaxonConcept ontology or "concept of a taxon concept", but in my view of the world every taxon asserted by a taxonomic authority is a taxon concept (rather than a taxon). So that includes the taxa in ITIS, CoL, Tropicos, etc.

I.e., isn't a taxon concept anything that a person or a group of people asserts to represent a taxon compliant with the applicable taxonomic code? 

-hilmar

Hilmar Lapp

unread,
Feb 26, 2013, 10:33:30 AM2/26/13
to tdwg...@googlegroups.com, J.Ke...@napier.ac.uk, Rob Guralnick

On Feb 25, 2013, at 11:29 PM, Bob Morris wrote:

> To whatever extent VoMaG cares about such provenance and standardizing it for the
> TDWG community, it would need \some/ standard predicates to address provenance, and maybe needs something as heavy as PROV anyway.

For one, I think VoMaG absolutely has to care about provenance. And personally if anything I find PROV too light rather than being heavy.

But I would also caution against overzealously turning every triple that might possibly have provenance (which don't, honestly?) into a full blown object. That just turns RDF onto its head.

The PROV model gives a good example for how you can have both (i.e., simple property, and full blown objects when you really need them), and at least in OWL retain full reasoning power by using property chains.

Also, while it's tempting to view everything we do as annotations and try to mold them into the OA framework, not everything is a nail even if you've got a wonderful hammer. Sure, you can hammer nearly everything, but that doesn't guarantee a delectable outcome.

Steve Baskauf

unread,
Feb 26, 2013, 1:52:27 PM2/26/13
to Hilmar Lapp, tdwg...@googlegroups.com, Rob Guralnick
I think it was pretty clear from
http://code.google.com/p/tdwg-rdf/wiki/TCSthread
and many previous discussions on the tdwg-content email list that unless it is defined very carefully, there is going to be a lot of disagreement about what a "taxon concept" or a "taxon" represents.  I think that a secondary benefit of keeping the TDWG TaxonConcept ontology is that it provides a fairly well-defined class, tc:TaxonConcept which can be "understood" by examining the properties (listed in the TaxonConcept ontology) that its instances can have.  That is a better circumstance than the class dwc:Taxon, whose definition is somewhat ambiguous and all-encompassing. 

Steve

greg whitbread

unread,
Feb 27, 2013, 9:04:38 AM2/27/13
to tdwg...@googlegroups.com, Rob Guralnick, J.Ke...@napier.ac.uk

Steve,

I would not use APNI as a reason for maintaining the TDWG ontology.  APNI is not an implementation of the concept ontology ... though the TCS, on which the TDWG ontology is based, looks a bit like the original APNI data model.  

We do have services delivering linked-data using the ontologies RDF and TCS XML but these are  intended only as a trial of these technologies.

Our contract to aggregate the major Australian, nomenclatural and taxonomic data sets within the NSL (National Species List) called for an implementation of the TDWG LSID policy which in turn recommends the use of the RDF vocabularies for the LSID getmetadata() response.  As the LSIDvoc evolved from the TDWG ontology ( or v.v.)  it also made sense - so as not to confuse - to use it to deliver the RDF resource for our linked-data experiments.    

The real task was to understand and aggregate these data to deliver persistent and reusable nomenclatural and taxonomic objects for the linking up and discovery of  information among name based, biodiversity data sets.  

We do not model taxa directly.  The exercise is simply to capture where names are used and the context in which they appear, along with necessary  assertions, links and references that might be used to better circumscribe them .  These are simple facts which we should be able to open to community management.  If we can provide persistent identifiers for these facts and dereferencing services for their web resources it becomes possible for others to both reuse our work and contribute back through data linkages.  As Hilmar states, every instance is potentially useful.

But to make it work we do need a better RDF representation.

From the NSL perspective there are a couple of issues with use of tc:hasRelationship that need to be addressed:

.   Throwing all relationships together is not a good idea.  It complicates everything. We manage these data as separate nomenclatural, taxonomic, classification and profile objects adding instances for relationship entities such as synonyms, nodes and assertions.  They all have relationship properties and all can be the target of an annotation.   

.  Our consumers mostly want what we have, so they can reuse these objects  to produce professional, value added, taxonomic and nomenclatural products.  This means extensible vocabularies that  accurately describe both the meaning and direction of these relationships. The generic approach used by the TDWG ontology makes it impossible to round-trip content  without loss of meaning.  Unhappy curators, unhappy clients.


greg


If you have received this transmission in error please notify us immediately by return e-mail and delete all copies. If this e-mail or any attachments have been sent to you in error, that error does not constitute waiver of any confidentiality, privilege or copyright in respect of information in the e-mail or attachments. Please consider the environment before printing this email.

--
You received this message because you are subscribed to the Google Groups "TDWG RDF/OWL Task Group" group.
To unsubscribe from this group and stop receiving emails from it, send an email to tdwg-rdf+u...@googlegroups.com.
For more options, visit https://groups.google.com/groups/opt_out.
 
 



--

Steve Baskauf

unread,
Feb 28, 2013, 1:21:12 PM2/28/13
to tdwg...@googlegroups.com
Greg,

Thanks for your very thoughtful and informative response.  When I said "Is there a reason to keep the TaxonConcept ontology?  ... it is being used by at least one major provider (APNI)" what I meant was that I think it would be bad to  "get rid of" a vocabulary whose terms are actually being used by a metadata provider and risk "breaking" something.  That's in contrast to something like, say, the TDWG DigitalImage ontology (http://code.google.com/p/tdwg-ontology/source/browse/trunk/ontology/voc/DigitalImage.rdf) which is effectively useless and as far as I know hasn't been used by anybody.  It would be possible "keep" the TaxonConcept ontology in place without necessarily recommending it for widespread use.

Because I'm not qualified to make a judgment about how effectively the TaxonConcept ontology does what it was set up to do, I'm glad to hear an opinion from you about its deficiencies.  I think that the task of setting up an effective "ontology of everything" was too big of a task for TDWG to take on and was perhaps doomed to fail.  Having said that, there may be some pieces of the TDWG Ontology that do still have some use and others which would be better replaced with something else.  It sounds like tc:hasRelationship and its values probably need improvement or replacement.  On the other hand, I think that tc:accordingTo is a very useful term which probably could be used effectively.  It's also possible that tc:hasRelationship could be useful for simple transmission of metadata as RDF, but not so useful for more demanding tasks such as reasoning.  So I'm glad for the ongoing discussion that should help  give us some direction on where we need to go from here.

Steve

Mark Schildhauer

unread,
Feb 28, 2013, 2:46:53 PM2/28/13
to tdwg...@googlegroups.com, robert.g...@colorado.edu, Ben Adams
Hi Folks,

I've been following this discussion with great interest, but am a bit frustrated that we seem to be wandering around in a  broad KR problem space. 

Do we have a concrete set of "Use Cases" stowed away somewhere, that can help clarify and prioritize the specific capabilities we hope to achieve through our vocabulary development?  This might help resolve what we can provide via development and promotion of (TDWG-approved) RDF vocabularies, or whether we need more expressive KR's e.g in OWL (and potentially referencing AO or PROV).

It seems like we all acknowledge the need for some RDF-like reification.   Some have pointed out that this is problematic from a SPARQL and performance issue, others are hopeful these capabilities will approve, and many agree that we can handle these challenges somehow, "under the hood" (e.g. via quads or indexing).  But I've also seen what I think are proposals to mimic RDF reification in RDF (I think Hilmar pointed this out), or we throw up our hands and say we need  AO/PROV (where admittedly it would be nice to use W3C vocabs if relevant).

To help focus discussion on what is probably a multi-tiered set of logical needs, I'd suggest we frame discussion through a well-defined set of Use Cases.  This can help us concretize the specific semantic capabilities/axioms we need case-by-case, and better address how these can be achieved via RDF vocabularies, and when we need something more expressive. 

And then, perhaps we can build our vocabularies through some nicely constructed patterns that are captured on a Wiki somewhere?  I'm afraid I have trouble following our sometimes informal syntaxes, especially when trying to determine whether some pattern/vocab will really resolve some semantic issue.

If we further consider how to make these vocabularies compatible ( like OWL adds to RDFS), we might be able to divide&conquer our way forward.  For example,  simple LOD exposure of raw biodiversity data could be a great starting point (if it provides some useful capabilities), but more advanced challenges, e.g. expressing and querying taxon concepts, might need more advanced inferencing.  It seems premature to be committing ourselves at the outset to RDF or SKOS, until we know specifically what we hope to achieve.

Perhaps  a set of TDWG Vocabulary Use Cases does exist, and if so, could someone point it out to me?  I haven't seen it referenced in this past discussion, aside from some tangential references to Taxon Concepts...

thanks!
Mark Schildhauer

Steve Baskauf

unread,
Feb 28, 2013, 5:15:51 PM2/28/13
to tdwg...@googlegroups.com, robert.g...@colorado.edu, Ben Adams
Mark,
Joel Sachs put together this page:
http://code.google.com/p/tdwg-rdf/wiki/UseCases
which is a sort of grab-bag of use cases that were accumulated from various people.  I am not aware that anyone has made a concerted effort to organize them in terms of what they tell us in terms of the direction that we need to move. 

I think that what you suggest is a great idea.  However, it would have to be done by somebody other than me.  I see my role as co-convener of the RDF group as to keep people talking and to try to make a record of what (if anything) we've concluded from our discussions.  But I don't have the technical skills to do what you suggest, so somebody else will have to take that up.  There are people in the group who could do it and I think something like 20 people were given "committer" status on the Google Code site when we set it up.   So if anybody wants to dig in - go for it.  If you don't have committer status, I can add you to the list.

Steve

Nico Franz

unread,
Mar 1, 2013, 11:00:00 AM3/1/13
to tdwg...@googlegroups.com, robert.g...@colorado.edu, Ben Adams
Thank you, Mark & Steve:

   That's a neat list to my mind, though I also feel that it is improbably successful at omitting use cases that have "by expert taxonomists, for even better taxonomic semantics from the bottom up" as their central motivation. Can I ask (Pathos 101) - is there a hole at the very center? Can we add to the list?:

21. Semantically enhancing taxonomic and phylogenetic revisions for improved concept integration across succeeding classifications.

   I could flesh this out later (as deemed appropriate; yes I'd like committer status - thank you).

Best,

Nico

Steve Baskauf

unread,
Mar 1, 2013, 8:47:25 PM3/1/13
to tdwg...@googlegroups.com
A thread summary of this discussion has been added to the task group's wiki:
http://code.google.com/p/tdwg-rdf/wiki/OAThread

Steve

Bob Morris

unread,
Mar 2, 2013, 11:31:08 AM3/2/13
to tdwg...@googlegroups.com, Hilmar Lapp, Rob Guralnick
To add to (my) confusion, and apparently not mentioned in the cited
TCSthread, I note that
http://tdwg-ontology.googlecode.com/svn/trunk/ontology/voc/
seems to have two files:
TaxonConcept.rdf
TaxonConcept.owl

Both declare themselves to be OWL ontologies.

TaxonConcept.rdf does not define a class named Taxon.
TaxonConcept.owl does, and avows it owl:equivalentClass to Taxon.

After a brief glance at both, I would say that on namespace grounds
alone, it looks really dangerous to use them both, and that could be a
tall order in trying to integrate data expressed in one with data
expressed in the other. This might be all the more so if someone
using TaxonConcept.rdf used something else to express a relation
between tc:TaxonConcept and some xx:Taxon. Worst(?) of all if that
usage happens to make xx: be the tc: namespace based on a misguided
notion that it is harmless to do so because TaxonConcept.rdf has no
Taxon class defined.

Bob
p.s. This has nothing to do with biological discussions about whether
"Taxon" means the same thing as "TaxonConcept". It's purely about
namespace issues and their consequences.




On Tue, Feb 26, 2013 at 1:52 PM, Steve Baskauf

Steve Baskauf

unread,
Mar 2, 2013, 2:02:00 PM3/2/13
to tdwg...@googlegroups.com, Hilmar Lapp, Rob Guralnick
Hmmm.  I had discovered that there were two of those files somewhere about 2/3 of the way through that thread and thought it was mentioned in there somewhere. But maybe not.  I believe (from memory) that there are several other differences  besides the absence of tc:Taxon in TaxonConcept.rdf:

- TaxonConcept.rdf has additional terms not found in TaxonConcept.owl
- TaxonConcept.rdf asserts that tc:TaxonConcept is a subclass of BaseThing from base TDWG ontology, while TaxonConcept.owl doesn't make this assertion.

Both files list identical last modified times and version numbers, so it is difficult to ascertain which one is the most recent version, or the "real" version.  However, after investigating which one is served when the terms are dereferenced with a request for content-type:application/rdf+xml, I concluded that the TaxonConcept.rdf is the "official" version.  So I have stopped referring to tc:Taxon and now refer exclusively to tc:TaxonConcept which I think is better anyway because despite the many thoughts about what a" taxon concept" is, I think people are closer to agreeing on a definition of what that is than they are to agreeing what a "taxon" is (maybe).  I've also tried to purge pages like
http://code.google.com/p/tdwg-rdf/wiki/Beginners#0.3.7._Biodiversity-related_and_General_vocabularies_and_ontolog
and
http://code.google.com/p/tdwg-rdf/wiki/TCSBasedTaxon
of references to tc:Taxon and the TaxonConcept.owl file.  Ah yes, I mentioned the presence of multiple file versions at
http://community.gbif.org/pg/pages/view/29079/section-1-a-review-of-the-tdwg-ontologies

Oh, I forgot, it's worse than that.  There's http://rs.tdwg.org/ontology2/ and http://rs.tdwg.org/ontology/ in addition to http://code.google.com/p/tdwg-ontology/source/browse/trunk/ontology/

It has been typical for TDWG to be sloppy and leave confusing refuse online like this - just look at the TDWG wiki and http://www.tdwg.org/standards/449/ which downloads the wrong standard.  Hopefully the efforts of the VoMaG task group will succeed in tightening things up a little (no, actually a lot). 

Steve

Hilmar Lapp

unread,
Mar 3, 2013, 2:14:49 PM3/3/13
to tdwg...@googlegroups.com, Rob Guralnick
What about planning for a cleanup-a-thon/document-a-thon-type workshop at TDWG 2013 for a group of people to work through these and related issue in a concerted fashion for an afternoon, rather than the piecemeal approach we're practicing now that never seems to wrap anything up.

-hilmar

Bob Morris

unread,
Mar 3, 2013, 2:38:49 PM3/3/13
to tdwg...@googlegroups.com, Rob Guralnick
+1

Éamonn Ó Tuama [GBIF]

unread,
Mar 4, 2013, 4:28:55 AM3/4/13
to tdwg...@googlegroups.com, Rob Guralnick
++1

-----Original Message-----
From: tdwg...@googlegroups.com [mailto:tdwg...@googlegroups.com] On Behalf

Steve Baskauf

unread,
Mar 4, 2013, 10:03:00 AM3/4/13
to tdwg...@googlegroups.com
+1

I didn't know that there was even a piecemeal approach...

Can somebody who is going to be at TDWG 2013 volunteer to spearhead this?


Éamonn Ó Tuama [GBIF] wrote:
Reply all
Reply to author
Forward
0 new messages