Yesterday I received a file with 129 translations units, an addendum to recently completed translation of a manual. Looking at it I realised that more than 50% had already been translated with the original translation of a manual from the same customer. The translation of the manual had been done by a colleague and I had aligned the English original and the German translation just the week before and after aligning the two I had stored the memory in Total Recall.
I was pretty amazed that after setting the new file up as a new project only about 5 segments were shown as match in the Matchboard. I was forced to manually copy the translation from the translated manual into the corresponding segments of the new translation.
I had already noticed this with a translation for a Bluetooth device.
How can I improve the recognition of matches?
Joachim: ...and after aligning the two I had stored the memory in Total Recall.
This doesn't sound like your typical Total Recall/Recall Memory situation, but on the other hand, I wouldn't know what went wrong.
I'd import the aligned files in a Memory for Segments, assign a high priority to it, and in this case, I'd probably run Translate | Insert All Exact Matches first.
Moi: ...I wouldn't know what went wrong.
After importing the aligned files into the Total Recall database, you did run Total Recall | Recall Memory, didn't you?
Woorden, after aligning the English original and the German translation I clicked on "Store Memory in Total Recall", see image. Then this new job came in and I expected that the segments identical in the two English versions would be displayed in the Matchboard (see attached file). This was not the case and I am wondering why not. Until a few weeks ago I used Across as CAT and that was exactly the behaviour I was used to from Across. Okay, new CAT, behaviour will be different. But what am I doing wrong that the matches don't turn up? After all, that's the reason to use a CAT.
Joachim: But what am I doing wrong that the matches don't turn up?
This explains it all. Total Recall is a database that is not included in the Automatic Workflow. It used to be there only for blistering fast manual concordance searches for huge files (like the DGT).
TMX memories (TMs) are included in the Automatic Workflow.
Now some time ago, His Igorness invented a a way to "reduce" the data in the huge database to only segments relevant for the current project, and... save them as a TMX memory. Brilliant. But it looks like you forgot that step, or rather, didn't know about it.
On the other hand, if you work with small/manageable resources - and I bet your aligned file isn't too big - you should import that file in a (new) TM for segments, assigning it a high priority. That will work better and faster, until the file gets too big (think of 200k segment or more).
Woorden, will all jobs then automatically go into the same TMX file?
Joachim: will all jobs then automatically go into the same TMX file?
A job will "go" into an XLF file, a different one for each job, unless you do decide to re-use an existing XLF file.
You can add segments and fragments to memories (TMs) to as many TMs as you want, and you can select the TM you want to add to.
I use a "Big Mama" for each language combination. The Big Mama holds all segments I ever translated, and therefor I assigned it a low priority.
Thank you Woorden. I know built the .TMX file from the jobs I still had available in CafeTran. But are you sure that Total Recall isn't integrated in the automatic work flow? Because, as you can see from the attached screen capture, Total Recall does seem to work. To me it looks like something in my settings for Total Recall seems to be awry. I'm just not sure what to change in the settings to make it work better.
Joachim: But are you sure that Total Recall isn't integrated in the automatic work flow?
Because, as you can see from the attached screen capture, Total Recall does seem to work.
That is because you used "Recall Memory" feature that creates a TM (*.tmx file), the Total Recall._TM in your screenshot.
The terminology used by Igor is the confusing part. The menu item "Total Recall" used to be called "External Database." That's what it is, a database. Then Igor introduced the feature of creating a TM (*.tmx) of only the segments relevant for the current project, and called the feature "Recall Segments." That's what it does, it recalls segments. For unknown reasons, that was changed into "Recall Memory," and that's confusing, because it doesn't recall a memory, it creates a memory consisting of the relevant segments for the project.
To me it looks like something in my settings for Total Recall seems to be awry. I'm just not sure what to change in the settings to make it work better.
The settings of Total Recall don't matter much. You should change the settings for the TM that Recall Memory creates.
Did you have a look at the Knowledge Base, unfortunately called Solutions? (Igor and terminology...)
Yes I read through all 6 articles dealing with Translation Memories. I know he uses SQLite as database, not a bad choice. MySQL is offered optionally for really large memories. At present my SQLite has exceeded 200 MB, so it's not that large.
I like the approach of the central repository in a database. What I don't understand is to fine tune it that it provides relevant segments. As I said in the beginning, more than 50% of a new document had been translated in another document just a couple of days before and also had been sent to Total Recall. I would expect these translations to come up and don't get it why they don't come up.
How do I change the settings for the TM that Recall Memory creates?
Sorry, I'll have to give up. Let that Kmitowski take over. F reshdesk stinks.
Thank you very much.
Woorden, in https://cafetran.freshdesk.com/support/solutions there is one more block of instructions "Working with Total Recall". I worked myself through these too now. To sum it up, it looks like the database (Total Recall) can substitute the "Big Mama" as you put it. I followed the suggestions there and lifted the size restrictions for the TM which is created at the time you start a new project. It then loads the whole database in memory, just like your "Big Mama", several hundred thousand segments in my case. Now I'm waiting for a larger job where I know that some segments should be in there to see whether it then works satisfactorily or not.
(I'm sick and tired of F reshdesk, but anyway):
Joachim: To sum it up, it looks like the database (Total Recall) can substitute the "Big Mama" as you put it.
Yep. However, the BM offers more metadata (like "creationdate") that can be useful.
It then loads the whole database in memory, just like your "Big Mama", several hundred thousand segments in my case.
No, it doesn't. I hope it doesn't. If you open the database table, it looks like this, the table:
It's there instantly, shows all entries, and can be searched.
If you "Recall Memory" (Recall Segments), you'll see this, a TM:
This can take a few seconds before it shows up, and it shows only segments relevant to your current job. Since it is a TMX file, is allows for fuzziness.
Now I'm waiting for a larger job
It should be useful for any job. In fact, I'd suggest you create a fake job consisting of one segment, and run Recall Memory (Segments).