The post-editing analysis in Memsource Cloud extends the traditional translation memory analysis to also include machine translation (MT) and non-translatables (NT).
When a user clicks in a segment, the current translation memory, machine translation, and/or non-translatable match gets saved for that segment and is later used to calculate the post-editing analysis.
The post-editing analysis is based on the target, therefore it must be launched after the post-editing job has been completed.
Analyze TM post-editing
The Analyze TM post-editing option is used for calculating the post-editing effort for matches from the translation memory (TM).
Analysis with Analyze TM post-editing enabled:
- Intended for a low-quality TM that can contain high percent matches that need to be heavily edited by the linguist.
- Analysis will show the post-editing effort for the translation memory.
- There are only 100% matches in the analysis (in-context 101% matches from TM have no effect on the calculation).
- A 100% match from the TM can become 0% if completely edited by the linguist.
- Any match from the TM can become a 100% match if accepted by the linguist without any change.
Analysis with Analyze TM post-editing disabled:
- Intended for a high-quality TM where the matches should be edited as little as possible, to keep the cost low.
- The analysis will show 101% and 100% (similar to Default analysis)
- The analysis will show TM matches offered to the linguist when the segment opened (not the actual linguist's post-editing effort)
- The analysis will show the post-editing effort for machine translation and non-translatables.
Machine translation and Non-translatables
Similarly to the Analyze TM post-editing settings, there are options to Analyze NT post-editing and to Analyze MT post-editing. Based on these settings, the MT and NT results can show the post-editing effort for each segment. It reflects the changes made by the linguist compared to the MT and NT suggestions.
- If the MT or NT suggestion was accepted without further editing (the linguist did not need to change it at all), it would come up as a 100% match in the analysis.
- If, on the other hand, the linguist changes the MT, the match rate will be lower. The score counting algorithm is identical to the one that we use to calculate the score of translation memory fuzzy matches.
- For NTs, any editing will cause the segment in question to become 0-49% NT.