CafeTran

The feature to Merge alternative translations is a great feature (and absolutely unique in CAT  land).

Is there any chance that we'll get such a great feature for the source side as well?

Merge source terms, would be a good name.

Example

I have these two term pairs in a glossary:

abgewinkelte	omgebogen
abgewinkelte;abgewinkelter;abgewinkelten;abgewinkeltem	haakse

Running Merge source terms would result in:

abgewinkelte;abgewinkelter;abgewinkelten;abgewinkeltem	haakse;omgebogen

This is the situation where a source term also occurs in another term pair.

There's also the situation where two related source terms share the same target term:

Verschlussschraube  afsluitplug
Verschluss-Schraube  afsluitplug

 These could be merged without any problem. However, things start to get tricky when source terms aren't meer spelling variants but refer to different concepts:

Laster;Lkw;LKW;Lastkraftwagen  vrachtwagen;vrachtauto
Laster  ondeugd;zonde;slechte gewoonte

 Of course you cannot merge both concepts (lorry/truck and vice).

Well, one can, since it's the responsibility of the translator to pick the correct translation. But it's tricky.

So what would be a good algorithm for this?

Actually situation 1 is exactly the same as situation 2, at least this would be for CafeTran Espresso. Just as 'abgewinkelte' is part of another source term nest, 'Laster' is also part of another source term nest. And in both situations, the target terms differ.

Duh, complicated. Who has a brainwave?

Another thought: Hunspell could match source terms of the type:

abgewinkelte	omgebogen
abgewinkelte;abgewinkelter;abgewinkelten;abgewinkeltem	haakse