W3C home > Mailing lists > Public > public-lod@w3.org > June 2010

Re: Request for Feedback, Suggestions on TaxonConcept Species Concepts

From: Peter DeVries <pete.devries@gmail.com>
Date: Sun, 6 Jun 2010 05:49:26 -0500
Message-ID: <AANLkTinkd2WSWTJJfF_AUAQ3N3OGg1Yrd1lfr9AJjgvJ@mail.gmail.com>
To: Toby Inkster <tai@g5n.co.uk>
Cc: public-lod@w3.org, pedantic-web@googlegroups.com
Hi Toby,

I have what I call a lightweight phylogeny that should match up with the
fields for the DarwinCore.

However, these are more for user convenience, since I don't believe that a
species concepts should be modeled as having a "fixed" phylogeny.

A phylogeny is an hypothesis which changes over time. The TaxonConcept
species concepts are modeled so that a species can have several phylogenies.
You will note that I link out to two alternatives at the class level for
each concept.

If the full phylogeny determines the species concept then *Felis concolor*and
*Puma concolor* are different things.

Same would be said for *Ochlerotatus triseriatus* and *Aedes triseriatus*.

Why not use the DarwinCore:

1) At the time I was doing this the DarwinCore was changing quite often.
    I tried to keep track so that if someone wanted to merge/translate it
would be fairly easy.
    This is why I have names that track their names but in a different
ontology.

2) There appears to be a subtile difference between how species are thought
of between DarwinCore and what I am doing.
    This has been changing over time but years ago when I was started it was
that the species name is the concept.
    Like before, this makes Ochlerotatus triseriatus and Aedes triseriatus
two different things.
    It seems that the DarwinCore team maybe changing their mind on this as
well.
    It seems that they are moving toward something that involves the species
concept being some combination of all the names that have been used.

To use the DarwinCore ontology directly and appropriately then you have to
accept how they see species and how that species relates to all the other
parts of the data set. I had advocated for a URI for "SpeciesConcept". It
was in DarwinCore then it was out, and I am not sure where it is now.
They seem to want to have the species concept be defined by referencing a
particular published description.

This is fine but they are assuming that everyone who tags a species
observation are doing so after reading the published description. I can say
that this might be how they believe or want to people people are operating
but that is not in my experience what they do in real life. By doing this
they are basically saying that the person who identified the species did so
based on the written species description. This is not true for the vast
majority of observation records. This comes into play when I think about,
did they mean Aedes triseriatus or Ochlerotatus triseriatus?

The last time I tried marking up records in DarwinCore I was not able to
make the kinds of SPARQL queries I am interested in. For some reason they
also insist on creating new vocabularies for widely used things like
latitude and longitude. Why not the geo vocab or some extension of it? Why
so many literals that would make more sense represented as URI's?

I think that some of these issues will be worked out in the future. And I
think that when full DarwinCore records start appearing on the LOD, they
will notice
some of these on their own. Fortunately, they seemed to have come around on
their previous insistence in using LSIDs, rather than Linked Data.

That said, if your vocabulary does not entail some compatibility issues I
would be happy to map these as equivalent properties.

This leads to a general question for the LOD community.

There are several versions of these taxonomies consisting of some
combination of kingdom, phylum, class, order, family, genus, specific
epithet and so on. Some groups use additional levels like subclass,
subfamily, etc. These are used for literals in DarwinCore, DBpedia, Bio2RDF
etc.

The problem that I run into when using another vocabulary you entail all
sorts of other modeling differences that lead to incompatibilities.

How would one map, these together without pulling in all the other
assertions included with their model?

As a literal a number of these have something similar to ontology:kingdom =>
"Animalia"

Which should be interpreted as this is a Biological Kingdom with the name
Animalia, however each of the vocabularies bring with them all sorts of
other statements about what a kingdom is an how it relates to other
entities.

For instance, Bio2RDF follows the subclassing of the NCBI Hierarchy. (Not
the wrong thing to do, just not the same thing)

This is one of the reasons that I don't use sameAs to link to related things
like Bio2RDF.

That said I should make point to reference your work and will do so.

Also check this:

curl -v -H "Accept: application/rdf+xml" http://purl.org/NET/biol/ns#kingdom

It should work like this

curl -v -H "Accept: application/rdf+xml"
http://lod.taxonconcept.org/ontology/txn.owl#kingdom

Note, the DarwinCore has the same problem.

I hope this has addressed you questions and concerns.

Respectfully,

- Pete







On Sun, Jun 6, 2010 at 4:18 AM, Toby Inkster <tai@g5n.co.uk> wrote:

> On Fri, 4 Jun 2010 18:06:08 -0500
> Peter DeVries <pete.devries@gmail.com> wrote:
>
> > I would appreciate feedback on these models and any suggestions for
> > how they could be improved. :-)
>
> The following:
>
> @prefix txn: <http://lod.taxonconcept.org/ontology/txn.owl#> .
>
> <http://lod.taxonconcept.org/ses/mCcSp#Species>
>         a owl:Class ;
>        txn:kingdom "Animalia" ;
>        txn:phylum "Arthropoda" ;
>        txn:class "Insecta" ;
>        txn:order "Lepidoptera" ;
>        txn:family "Nymphalidae" ;
>        txn:genus "Danaus" ;
>        txn:epithet "plexippus" ;
>        txn:author_year "(Linnaeus, 1758)" ;
>        txn:commonName "Monarch Butterfly" ;
>        foaf:page <foo> .
>
> Appears to be an almost identical way of representing a species concept
> to the one I came up with a couple of years ago:
>
> @prefix txn: <http://purl.org/NET/biol/ns#> .
>
> <http://lod.taxonconcept.org/ses/mCcSp#Species>
>         a txn:Taxonomy ;
>        # note: txn:Taxonomy rdfs:subClassOf rdfs:Class.
>        txn:kingdom "Animalia" ;
>        txn:phylum "Arthropoda" ;
>        txn:class "Insecta" ;
>        txn:order "Lepidoptera" ;
>        txn:family "Nymphalidae" ;
>        txn:genus "Danaus" ;
>        txn:species "plexippus" ;
>        txn:authority "Linnaeus, 1758" ;
>        txn:commonName "Monarch Butterfly" ;
>        txn:seeAlso <foo> .
>
> I wonder if you could reuse the URIs I minted rather than creating new
> ones?
>
> If we could stamp out any incompatibilities between the two ontologies,
> then I'd be happy to point the http://purl.org/NET/biol/ns URI at your
> ontology, so we'd just have a single merged ontology.
>
> --
> Toby A Inkster
> <mailto:mail@tobyinkster.co.uk>
> <http://tobyinkster.co.uk>
>
>


-- 
----------------------------------------------------------------
Pete DeVries
Department of Entomology
University of Wisconsin - Madison
445 Russell Laboratories
1630 Linden Drive
Madison, WI 53706
GeoSpecies Knowledge Base
About the GeoSpecies Knowledge Base
------------------------------------------------------------
Received on Sunday, 6 June 2010 10:50:01 UTC

This archive was generated by hypermail 2.3.1 : Sunday, 31 March 2013 14:24:27 UTC