Let's take taxonomy run by NCBI.
Approach 1.
.ence- pseudoprefix for id of a taxon provides us with around than 800 000 taxa.
.encesoxanoxa is a live being of species Homo sapiens (id 9606).
.encesomusoze is Bonobo earlier known in Lojban as jbonobo.
Definitely, {.encesoxabimu} (Felis catus, id 9685) is easier to use than {zdani mlatu}.
jbonobo happily got a new name, jboremna can start using it.
Approach 2.
Another approach is to make an algorithm where Latin names are adapted into Lojban morphology so that back conversion is unambiguous.
We've successfully converted >99% of names of species.
You can see a table with all those names and a special columns telling whether back conversion from Lojban correctly restores original Latin name. We've ignored distinction between upper and lower case since it's of no importance:
Few names haven't been converted:
Opisthoteuthis sp. B-PCHH2001
Gautieria sp. HH2221Swiss
and only 2,600 other names out of ~800,000. Those are mostly names of subspecies (that can be additionally handled when capital letter abbreviations are found) or ad hoc names that are better to quote with zoi. ... zoi. as provided by Lojban language.
la jbovlaste is the main database of Lojban words and soon it will get 800,000 new words.
</aftershock>