Hi Susanna R Miles 

Nagoya is mostly correct, but just to clarify.

Susanna R Miles said:

I know Access databases usually don't function well if over 2 GB. - Is this an issue for the .sdltm file? - and if yes, what do you suggest I do.

Studio uses SQLite for Translation Memories and these in theory can handle up to 140 TB, so 5GB isn't an issue.  You can read more if you're interetsed here:

https://www.sqlite.org/limits.html

Susanna R Miles said:

Numbers, even part numbers (in letter form) and tags apparently aren't recognized as differences.

This should reduce the size of your TM.  Recognized numbers don't need to be duplicated since Studio sees then all as one and "should" be able to autosubstitute the correct value without storing every single number you ever translate.

Susanna R Miles said:

2) I know I can export to .tmx and then re-import into a new TM.

If I do that, what do you recommend for settings to strip out duplicates, or TUs that maybe differ only by tags

The default settings would strip out true dupicates.  If the difference was only tags then you should probably leave them in there as these are not duplicates.  If you remove the TUs with tags then you could get a 99% match instead of a 100% for a document you have translated before with tags.

You could also have a play with this application which has a feature for duplicate removal:

https://appstore.sdl.com/language/app/sdl-translation-memory-management-utility/131/

But make sure you back up your TM before you start... I'd recommend you make a back up to TMX as well and do these backups regularly if you don't already.

Hi Susanna R Miles 

Nagoya is mostly correct, but just to clarify.

Susanna R Miles said:

I know Access databases usually don't function well if over 2 GB. - Is this an issue for the .sdltm file? - and if yes, what do you suggest I do.

Studio uses SQLite for Translation Memories and these in theory can handle up to 140 TB, so 5GB isn't an issue.  You can read more if you're interetsed here:

https://www.sqlite.org/limits.html

Susanna R Miles said:

Numbers, even part numbers (in letter form) and tags apparently aren't recognized as differences.

This should reduce the size of your TM.  Recognized numbers don't need to be duplicated since Studio sees then all as one and "should" be able to autosubstitute the correct value without storing every single number you ever translate.

Susanna R Miles said:

2) I know I can export to .tmx and then re-import into a new TM.

If I do that, what do you recommend for settings to strip out duplicates, or TUs that maybe differ only by tags

The default settings would strip out true dupicates.  If the difference was only tags then you should probably leave them in there as these are not duplicates.  If you remove the TUs with tags then you could get a 99% match instead of a 100% for a document you have translated before with tags.

You could also have a play with this application which has a feature for duplicate removal:

https://appstore.sdl.com/language/app/sdl-translation-memory-management-utility/131/

But make sure you back up your TM before you start... I'd recommend you make a back up to TMX as well and do these backups regularly if you don't already.