Databases for multiple view taxonomy

David Remsen dremsen at MBL.EDU
Wed Jan 29 19:03:59 CST 2003

Hi Anthony

We are building a system right now.  The taxonomic name server that Im
assembling as we speak (instead of eating dinner) is essentially a
networked database with several different ways of using it.

I am starting to put together some pilots for developing interfaces
over the course of the next months.  I would very much like access to a
'real' piece of monographic work.  Do you a sample document as well as
specifics about what you would actually want from an external system.

The current pilots involve full-text XML documents from AMNH and NYBG
which Im going to try to develop some generic XML parsing tools around.
  Essentially I want to be able to point an XML document at our name
server, have it identify taxa within the document based upon the
multi-class data model, iterate through which taxa and treatments I
ascribe to, and then take all the relevant name and classification
information from the server and give me back a supplemental XML
document with pointers back to the source XML.  I have XML with
specific name-related tags as well as a case where name information is
contained within freetext.

--- cut here if you don't need to hear any more but what follows is a
bit more background.

The model for how it manages different classifications is based on
several sources.  The primary model is the Metathesaurus of the Unified
Medical Language System at the National Library of Medicine.  This
system is a complication of multiple medical terminologies and the
relationships between concepts within these terminologies.  Other big
influences have been folks like Richard Pyle from this list, MoreTax
and Nomencurator.  It is a first step in exploring how best to deal
with a difficult issue but we have enough of a model that we are
putting together a system and then some prototype applications to use
it.   What I have learned is that there are lots of different
stakeholders with different needs.  I know what my needs and dream
system would be but it might be quite different from yours.

The salient points:

The Taxonomic Name Server views all taxonomic concepts as, in Richard
Pyle's terms, "assertions."   A taxonomic concept is someone's opinion
of a particular taxon based on an original description.  These concepts
or assertions have relationships both within a specific taxonomic
treatment and without, with other treatments.  These relationships can
be hierarchical or set-related.  Everything is unique and ascribed to
some source authority.  I make no assumptions.   I am putting all sorts
of interesting sources in such as 12 different classifications for a
single species (Cercopithecus mitis, a monkey).
Each instance of the C.mitis concept is unique with further rows then
ascribing relations between them.  They might all be the same thing but
that is left for the source authority to "assert."  Im currently
loading this database with approx. 1 million taxa and 1.5 million name
strings so we can see what it does at scale.

We will talk to the name server primarily through an XML-RPC or a SOAP
interface.  These two systems are kissing cousins and I prefer XML-RPC
to SOAP but we have functions for both.   This interface defines
functions on the server which are called from the client side to
deliver this taxonomic 'metadata' to the client.

Our initial prototype applications will fall into a few categories
which I think account for much of the usage of this system.  I'm in the
archive/library/data business so our needs might not be your needs but
it's a start and I'd like to hear about what people would use a
'system' for.

Our pilot applications will explore:

Full-text XML parsing - specific name tags plus names within freetext
RDBMS applications and Marc record systems
Dictionary-type freetext parsing

For much of our needs we are interested in using a dynamic taxonomic
authority to help us index the contents of an information source and
then dynamically import and integrate supplemental name and
classification information and repeat this process dynamically.

My goal is to have a workable API and a suite of client tools in Java,
PHP, and Perl by April with demonstration applications using it.  What
I expect at that point is to attack it and determine if it does what it


On Wednesday, January 29, 2003, at 06:04 PM, Anthony Pigott wrote:

> Dear All
> Some while ago there was some discussion about database systems /
> software to support multiple taxonomies or publication centred
> taxonomy,
> i.e., all the past views rather than one current 'accepted taxon' view.
> There are a few systems around with at least prototype software
> available, e.g. Syngraph, Nomencurator and Prometheus.
> I'd be grateful to hear if anyone has experience of using any of these
> or similar systems in a 'real' piece of monographic work.  I'd like to
> use such a system and whilst very tempted to do the modelling and
> database design myself, I know better than to try if there's something
> usable out there. Thanks.
> Regards
> Anthony
David Remsen
uBio Project Developer
Marine Biological Laboratory
Woods Hole, MA 02543

David Remsen
uBio Project Developer
Marine Biological Laboratory
Woods Hole, MA 02543

More information about the Taxacom mailing list