Hi Tyler Snow 

This is quite an interesting one and I'm not sure what the right answer is as both 2014 and 2017 look wrong to me.  2014 counted it correctly because it behaved incorrectly, and 2017 introduces a benefit on one hand and takes it away with the other.

So, 2014 does not recognise these as numbers at all.  It sees them as words.  I don't know if this is correct to write like this:

제1 meaning 1st

제2 meaning 2nd

제3 meaning 3rd

제4 meaning 4th

So please correct me if this is wrong, but I used this as a test.  In 2014 these are 4 words and analysed like this:

Each one is handled as a word because the numbers are not recognised in this context.  It you translated these would would end up with four TUs in your TM.

In 2017 the analysis is this:

This time you end up with one TU in your TM because the numbers are recognised as separate tokens.  So you gain with interactive translation since these forms will be translated automatically no matter what the number is, but the analysis has doubled.

So it looks to me as though the problem really is that '제1' and others should be seen as a single token and then the analysis would be four words and you would still only have one TU in your TM.

I'll pass this over to development for review like this if you are in agreement that this is how these numbers should be handled?

Regards

Paul