unique numbers for species

Markku Savela msa at BURP.TKV.ASDF.ORG
Thu Oct 11 23:12:10 CDT 2001


> I already outlined one of the many potential problems of trying to map
> multi-part textual data across databases ("Smith & Jones", "Smith and
> Jones", etc.).

This I briefly touched, when I said the "id"'s should be
"normalized". It would mean specifying some simple "transcription
rules", so that "Smith & Jones" and other fuzzy things would always be
written the same way. Additional fuzzines is brought by the fact, that
different views about older dates exists.

> To a computer, one exception is all that's needed to destroy an
> otherwise elegant system

Yes, but like any system, even one based on central registry, must
have ways of handling occasional error of duplicating id and alert the
person maintaining the database.

> Since we're talking about expanding the practical effectiveness of computers
> as tools to help us do our work more efficiently and effectively, we should
> be thinking in terms of what makes sense to a computer, rather than what
> makes sense to a human.

I'm actually talking from computer background, and computers are
ideally suited to handle strings too, not just numbers.

I'm not trying to force anyone to follow my views. I'm just trying to
open eyes to possibilities that exist. This should be worth exploring.

It is just question of designing interchange formats, that is
independent of the local internal numbering, and that is not a big
problem.




More information about the Taxacom mailing list