We are often faced with the situation where a legacy or unclean TM needs to be applied to a workflow in order to review or correct the content due to global errors, legacy quality, change of language rules (terminology, style, etc.).
Within one TM sequence, you would apply two TMs:
- Clean TM: content has been through review so matches from this TM are up to date
- Unclean TM: content has not yet been through review so matches from this TM are not up to date
Currently, you have the option to lock or keep unlocked 100% matches but you cannot differentiate between TMs. The only way to do this is to apply a penalty. Penalties reduce the match from a 100% that requires review to a 99% fuzzy match that requires review.
There are two issues with this:
- Translators cannot distinguish easily between 99% matches that have content changes and 99% matches that just need review. This increases the chance that a translator will under-edit a genuine 99% match.
- Changing the cost band from a 100% match to a fuzzy match has cost implications as the ppw of a fuzzy match is higher than a 100% match. This leads to a higher cost to the client or a higher cost paid to external resources. In these cases, TMS can be updated to separate the 99% fuzzy band and a ppw negotiation needs to be done between all parties. Changing fuzzy bands also has an impact on reporting data.
Offline, it is possible to apply TMs in a sequence and lock segments from one TM before applying the next TM. The analysis can separate locked segments from unlocked segments to word counts and effort is analysed accurately.
Is it possible to replicate this feature in an online environment too?