1. As preliminary matching goes on, matched TUs in the original TM are marked as such (or removed from the original TM (in RAM only)). 2. When it is finished, CT can export the original TM with these marks (or the reduced original TM integrated with the preliminary matching memory). 3. Then, the exported memory consists of two groups of TUs. (i) TUs that are preliminary matches: For them, the memory matching type is set as Automatic. (ii) Other TUs: For them, the type is set as Manual (dormant) so that they can be searched whenever the user wishes to do a context-independent search or check below-fuzzy TUs.
Then, there would be a far less trade-off between speed and coverage.
1 person likes this idea
over 2 years ago
As preliminary matching goes on, matched TUs in the original TM are marked as such.
Preliminary matching memory is created as implemented now.
Upon export, it is saved in a new folder.
At the same time, the marked TUs are removed from the original TM (to avoid duplication and reduce workload) and the rest is saved as a new TMX in the same folder.
The preliminary matching memory is set as read-only and Automatic, and the (reduced) original TM as read-only and Manual.
That folder is zipped, and its extension is changed to a unique one, like .pretmx.
When it is opened, CT recognizes the different settings of the two TMs, while presenting them as one to the user.