ITIS (an explanation of GBIF's data integration activities)

[Donald Hobern]

Richard Petit has raised several issues about the presentation of
taxonomic information by ITIS and GBIF.  Some of his concerns arose from
an e-mail he had received from me (sent on behalf of GBIF, not ITIS):



>> I then sent in four of five corrections and received a reply from

>> ITIS (will forward to Dr. Lane off server but will send to anyone

>> interested) in which I am advised that "it is important that the

>> search interfaces continue to list these names since a user may

>> recognise them as misspellings for names of taxa in which he or she

>> is interested."



It may be helpful for me to explain a little more about GBIF's data
integration activities.  I hope to be able to clarify our approach and
to remove some fears about what we hope to achieve.



GBIF aims to provide electronic access to key primary resources for
biodiversity informatics and to ensure that these resources are as
interoperable as possible for use by different communities.  To this
end, we have focused on two different levels.



First, we are encouraging the digitisation of biodiversity-related data
using well-defined data standards and access protocols.  Most of the
work in 2003/2004 has been in establishing a global network of specimen
and observation databases (building on the work of many regional and
thematic networks).  These databases are being accessed using TDWG
standards (the DiGIR protocol, with the Darwin Core record format, and
the BioCASe protocol using the ABCD schema).  We are also working with
providers of nomenclatural and taxonomic information (including ITIS and
Species 2000) to develop similar models to simplify access to resources
such as nomenclators, taxonomic revisions, regional checklists and other
significant species lists (e.g. red lists).



Secondly, we are developing central infrastructure to help users to
locate all of these data resources and to select those which address
their interests.  We have established a central registry with
descriptive and technical metadata concerning the resources which can be
accessed through the network.  During the next year, we will provide a
web service to allow users and software tools to search for these
resources using any of the fields defined in the registry database.  We
are also building an index to all of the taxon names in use in all of
these resources.  As of today this index holds information from the
ITIS/Species 2000 annual checklist and from the taxon names used in
specimen and observation records throughout the network.  This index is
not in any way regarded as an "authoritative source" of taxonomic
information.  It is rather a tool to assist users to discover which
resources include data on different taxa.



The current GBIF Data Portal (http://www.gbif.net/) is a prototype
intended simply to allow users to see the data offered today by data
providers within the GBIF Network.  Its search function offers a view
into the data index and therefore includes a wide variety of names which
may in many cases be invalid, misspelled or otherwise incorrect.



The Data Portal includes information from ITIS and Species 2000 and uses
it for two purposes.  The ITIS/Species 2000 taxonomic hierarchy is used
to allow users to browse much of the accessible data without searching
directly by name.  Synonymy data from ITIS/Species 2000 is used to help
users to find relevant specimen and observation records which may have
been provided using different names from the one used in a search
request.



All data presented through the GBIF Data Portal interface is credited to
the provider of the information.  We identify ITIS or one of the Species
2000 Global Species Databases as the source for each name which they
include.  We report their opinion on the acceptability of the name (e.g
"accepted name", "unambiguous synonym for Puma concolor").  For all
other names in the index (i.e. for all names which we only know from
specimen or observation records), we state that their position in the
overall taxonomic hierarchy is "tentative".  This approach is meant to
highlight that these names are of unknown quality, but we recognise that
this needs to be made much clearer.  We therefore intend to change the
interface to provide a clear separation between accepted names,
unaccepted names and names of unknown provenance.  As I stated in my
e-mail to Richard, we believe however that all of these names have to be
included in an interface which is intended to allow users to find data
regardless of the name under which it may have been shared.  This is
very different from an interface such as that offered by ITIS, which
does seek to provide an authoritative judgment about the names included.
GBIF will also offer interfaces to resources which provide authoritative
taxonomic judgments, and we need to ensure that users can easily
understand what sort of data they are searching and viewing at any
point.



As GBIF develops, we should see a wide range of different tools and
portals based on the underlying data resources.  These will serve the
needs of different types of users and will include access to different
taxonomic hierarchies.  Many different groups will be able to develop
the tools that meet their particular needs.  This will be made possible
because GBIF is using well-defined standards and protocols, and
encouraging the development of reusable, open-source software
components.



Many users will have no ability to choose between different taxonomic
views and will require the system to provide them with a default
taxonomy (something which will clearly be a great challenge in itself).
However there are many other users who will wish to use, or will need to
use, a particular taxonomy.  These taxonomies should be made available
through the network and will be used to offer different paths into the
overall pool of data.  Taxonomists should be able to organise their
personal view of the data using their preferred taxonomy (even using a
local working version of a taxonomy for this purpose).  Those working in
government departments or conservation organisations should be able to
access data using the species names required by their work (red lists,
regional checklists, etc.).  Any user should be able to enter the system
and browse just the names from IPNI, IndexFungorum, "Catalogue of the
Pseudoscorpionida" or "Distribution and Taxonomy of Birds of the World",
along with metadata describing the source for each list and its intended
use.  Many of these lists will be "authoritative" views for the groups
concerned.



I am fully aware of the enormous gap between where we are today and
where we hope to be in the near future.  I also recognise that we will
need access to much fuller information about the relationships between
different taxonomic concepts if we are going to be able to allow all of
the available data to be re-organised using a number of different
taxonomies.  These are the problems which GBIF is trying to address.  I
would follow Richard Pyle's plea on Monday and encourage you all to
contribute to this process, in particular to the development of the
various TDWG standards needed to make this work.  Of most relevance
right now is the development of an exchange standard for taxonomic name
and concept data (see http://tdwg.napier.ac.uk/phpwiki/index.php).



Donald



Donald Hobern (dhobern at gbif.org)

Programme Officer for Data Access and Database Interoperability

Global Biodiversity Information Facility Secretariat