Specimen identification
This is a beginning of specification for an application, that could be
a browser extension, realizing a Specimen identification. For now it doesn't
take into account the contribution of image recognition.
The user narrows its choice until there is only one species possible.
But the user has more freedom than in a tree-driven dialog à la
Delta, (s)he can enter at any time a character that appears particular
and discriminating. On the other hand, the classical identification keys
of Floras are tree-structured, and take in account only a kew characters.
There will be an input field for a character with three ways of entering
organ, property, and value:
-
automatic completion in an input field
-
point and click on an image of a schematic plant with all organs present
(with a possibility to supress organs that are not present)
-
pull-down menu
This is possible because the relevant meta-data has been downloaded
from server: list of organs, properties, and values (see XML
protocol).
At any stage the following items are recomputed and proposed to the
user:
-
the number of species corresponding to the list of characters entered
-
the most discriminating questions
-
the list of possible families or genus, possibly presented as a tree view
When the identification is advanced enough, the system will display an
identikit picture (portrait-robot) of the plant or flower or leaf, that
will be recomputed at each step.