oops and parsing taxonomic rank

Robert A. (Bob) Morris ram at CS.UMB.EDU
Mon Jun 25 22:24:17 CDT 2001


Apologies to the group for a message meant to be sent to Christian.
Most of it was previously sent to taxacom anyway...

To salvage the goof and continue along the question of context and learning:

1. I do not require full automation as one goal will be to develop tools
for biologists to critique some of the automatic markup of digitized
journals.

2. Suppose a system were built in which the rank were deduced from the
ICZN and ICBN and perhaps a few other rules, and that system produced
its rank assignment together with the sentence containing the
assignment. Would there be a high likelyhood that a suitably trained
human expert could correct the assignment from the context of a single
sentence?

3. If there were some indication of internal contradictions
(e.g. different rank assignment to the same name in the same
document), would this aid in the human evaluation?

4. Would a hypertext link be useful that allows the evaluator to jump to the source at
the point of usage. (Seems obvious that the answer is yes, but hey
good software engineers never impose their ideas on the clients...)

5. Should I take this discussion elsewhere?




More information about the Taxacom mailing list