computer aided identification

[Craeymeersch, Johan]

The Benthos Ecology Working Group of the International Council for the
Exploration of the Sea (ICES) discussed at their last meeting the
feasibility of producing computer-aided techniques to achieve
standardization of identification and faster update of keys.

The group made a list of desirable attributes for such identification
programs:
 - no restriction of the order in which the characters can be used
 - the program should advise on the most suitable characters for use at any
stage of an identification
 - the program should use probabilistic identification methods
 - the program should give an overview of all characters used
 - the program should be capable of restricting identification on a
geographical basis
 - the program should allow flexible display of illustrations
 - the program should include information on the species' distribution and
ecology, and important references
 - the program should keep track of the species identified before by that
person and warn if the species was new for that person
 - it should be possible to add notes to the species' information
 - the program should be easy to update.

At their meeting a list of programs (list produced by M. Dallwitz) was
available. The group agreed DOS-programs are mostly not user-friendly and
shouldn't be evaluated by members of the group.
Time was too short however to start any evaluation. The group want to do
this at their next meeting (May 1996, Aberdeen). The group also will
evaluate whether members of the group can easily produce (parts of) keys
themselves.
At present, I have (demonstration versions of) following programs:
 - INTKEY4 (M. Dallwitz, Canberra, Australia) (Windows)
 - IdenfiyIt (ETI, Amsterdam, the Netherlands) (Windows)
 - XID (XID services, Pullman, USA) (DOS, but Windows in the near future)

I would like to hear the pros and contras of the attributes mentioned above
(or those that should have been mentioned but are not). .

As a starter, a point made by Richard Old (XID) on the probabilistic
methods:
"This is counter to the entire philosophy behind the XID system.  I would be
glad to engage in a disscussion of the pros and cons of probabilistic
systems however suffice it to say that XID was created to provide a better
approach in response to such systems."

Johan Craeymeersch
Netherlands Institute of Ecology
Vierstraat 28
4401 EA Yerseke
the Netherlands
craeymer at nioo.nl
(in the near future :   craeymer at cemo.nioo.knaw.nl
bureaucracy, isn't it nice?)