I guess this can be considered a bug, not an idea for improvement... I've forwarded it to SDL...
Just an opinion...
Cheers
Edric
Subject:
Comment:
The contents of this post will automatically be included in the ticket generated. Please add any additional comments or explanation (optional)
Maarja Ora Estonia Local time: 08:59 English to Estonian + ...
Has this problem been sorted out?
Jun 23, 2009
[quote]Edric Barbosa Filho wrote:
This post is for people already experimenting with Studio 2009, Beta or RC.
I am currently trying Studio 2009 and could not find where I can configure memory imports with the options "leave unchanged", "overwrite" etc. for existing TUs.
Tried the filters, but could not find an appropriate one.
Has this problem been sorted out? Does anybody know? I would like to use Trados 2009 Freelance Suite since I think the AuttoSuggest Creator would be useful but I'm afraid of using Trados 2009 and importing my translation memories.
Subject:
Comment:
The contents of this post will automatically be included in the ticket generated. Please add any additional comments or explanation (optional)
Nope, Maarja, those options are still missing. One of the reasons that I keep on using my Suite "Sweet" Freelance 2007, keeping aside my Studio 2009 bought early.
I hope SDL is working hard on that...
Rgds
Edric
Subject:
Comment:
The contents of this post will automatically be included in the ticket generated. Please add any additional comments or explanation (optional)
Andrei Vybornov Russian Federation Local time: 10:59 Member (2008) English to Russian + ...
What is going on?
Nov 22, 2009
The idea is marked as implemented on the SDL site, but I cannot find this setting anywhere (I use Studio SP1). Can anybody explain what is going on?
This feature is critical for me too...
Subject:
Comment:
The contents of this post will automatically be included in the ticket generated. Please add any additional comments or explanation (optional)
Studio treats imports differently: "existing translations" for Studio is a TU with both source and target segments equal to the one imported. Bottom line, Studio always "merges" the imported segments, except perfect duplicates.
SDL gave a reasonable explanation for that: when importing, if the memory contains two TUs with same source but different targets, which one would be overwritten? That's a good point, since I believe Trados simply replaced the first one...
That behavior is a bit questionable when you have a memory with no duplicate source/different target TUs, a string resource file, for example, and you need to import a memory with updated terms: you'll end up with at least two segments, the original and the updated one.
But you're not 'left alone' on that: you simply do not import the new memory, but rather use it as the 'main memory' with penalty zero in the memories to be used in the translation (you can use as many as you wish), and applies a penalty to your 'unupdated memo', for example "2". When you translate, Studio will use the 100% matches of the new memo, because the old memo will have only 98% matches (100-2).
If you set Studio to "update" both memories during the translation, you'll end up with the 'old' updated too, using the updated terms of the new one. Besides, you can go to the "Memories" pane and 'batch-edit' the old memo with the terms you've just saw in the Editor.
For each client and/or subject I keep one "base memory", maintained by me, verified etc.: when the client sends a job with an updated memory (i.e. "software2009" to "software2010"), I use it as 0% penalty and include the "base memory" with a 2% or 4% penalty (just to see the TUs really penalized on TM pane). In the end I have my base memory updated just on what I really reviewed, not TUs that could have been translated by someone else (the "new" one), and keep the new one just in case of further jobs with the same memory.
It works great for me.
Rgds
Edric
[Edited at 2009-11-22 13:45 GMT]
Subject:
Comment:
The contents of this post will automatically be included in the ticket generated. Please add any additional comments or explanation (optional)
Andrei Vybornov Russian Federation Local time: 10:59 Member (2008) English to Russian + ...
This will lead to unnecessary multiplication of TMs
Nov 23, 2009
Hi, Edric
Thank you for your answer. I never really thought about overwriting one of the two duplicate TUs. Good point indeed. But I think the solution for this problem is not very good. After all there won't be too many duplicates which will need to be updated during import, therefore it would make more sense to me if Studio asked about each such TU in the process of importing, showing me both TUs and letting me decide.
The solution that you proposed can work, but will lead to unnecessary growth in the number of TMs that you will have to handle. May get lost after awhile. Especially if you have many customers and use Customer 1 TMs as reference for Customer 2.
MultiTerm allows to search for duplicates. I wonder why there is no such thing in Studio for TM maintenance. Could save the day...
What I really fail to understand is why the idea which is clearly about choosing between merging & overwriting has been marked 'accepted' on the Ideas.com, but has not been implemented. Looks like a bald-faced lie. Did SDL people explain that too?
Thank you for your answer anyway.
Best regards,
Andrei
Subject:
Comment:
The contents of this post will automatically be included in the ticket generated. Please add any additional comments or explanation (optional)
The "accepted" in the idea is because the idea was mine, SDL explained to me that new approach and I found it fine at the time, so it was accepted.
I agree with you that the process is not ideal the way it is now, and it really can lead to many duplicates just because there is a comma difference in the targets, for example.
The process can be really improved, from the user point of view.
Best regards
Edric
Subject:
Comment:
The contents of this post will automatically be included in the ticket generated. Please add any additional comments or explanation (optional)
Andrei Vybornov Russian Federation Local time: 10:59 Member (2008) English to Russian + ...
Could you please clarify?
Nov 23, 2009
Hi, Edric
Thank you for your prompt reply.
Having given it a second thought, I realized that keeping a “base memory” alongside with the “client’s memory” will bring us back to where we started.
I see how it works with your “base memory”. Keeping it clean and populated only with the TUs that you reviewed makes perfect sense. Cristal clear so far.
But there is something I am missing. What happens to the TMs which the client sends you (again and again)? You don’t use them for just one project and then throw away, or do you? You actually mentioned that you “keep the new one just in case of further jobs with the same memory”. So I am trying to imagine what you do when that new job comes (and not surprisingly with yet another updated TM). You have that “client’s memory” that you kept from the last job and the newly arrived one. Do you add them to the project as separate TMs? Then very soon you will be completely lost in your TMs, because you will have as many TMs as there are jobs. Or do you import the newly arrived TM into the one that you kept from the last job? If so, then here comes the issue of “merging vs. overwriting” again. Back where we started. You will have duplicate TUs and will not know which one is right.
I will be very grateful if you explain how you handle it. I am now in that painful process of converting from the good old Trados to Studio and the chances are I may skip Studio and go directly to MemoQ, although I do like the AutoSuggest feature and MultiTerm more than LCS and Terminology module in MemoQ. So any advice how to tame Studio will be appreciated.
PS: May be I did not make it clear in my previous post, I meant that SDL marked this idea as accepted on their site, included it in the list of ideas implemented in Studio, and yet there is nothing like it in Studio. The fact that you have learnt to live without proper import functionality does not excuse SDL people at all. They claim to have done what they have not done.
Best regards,
Andrei
Subject:
Comment:
The contents of this post will automatically be included in the ticket generated. Please add any additional comments or explanation (optional)
I just keep the "new client memory" until the client sends a newer one. When he does, I throw the previous away: my base memo is correctly updated, and I always see the latest terms. Of course, if the client says that the newest memory complements the previous one, I just add it to the list. If he says "use both" I penalize the older 1%, but I have never had more than 2 memos from the same client, so that scenario of having many memories is pretty difficult to happen... If I see 2 different translations for the same term, I use the newest and point that to the client. Besides, most of the time the client sends an updated termbase that I check...
But I do agree that the Import feature should be made better to avoid duplicates just because of a comma or a small word, for example, presenting a list for unambiguation in the case you do not want to use more than one memory. One of such cases is when you have to combine memories because an important one has less than 25,000 TUs and you cannot use it alone to generate an AutoSuggest dictionary...
But be easy on the SDL guys: they are working really hard and they check all these posts here. I wouldn't be surprised to see they telling us more about that importing feature, or even implementing that unambiguation list in the future...
Best regards
Edric
Subject:
Comment:
The contents of this post will automatically be included in the ticket generated. Please add any additional comments or explanation (optional)
Nicolás Valencia Colombia Local time: 01:59 Member (2007) English to Spanish + ...
A fair compromise
Oct 20, 2010
Edric Barbosa Filho wrote:
Hello,
Studio treats imports differently: "existing translations" for Studio is a TU with both source and target segments equal to the one imported. Bottom line, Studio always "merges" the imported segments, except perfect duplicates.
SDL gave a reasonable explanation for that: when importing, if the memory contains two TUs with same source but different targets, which one would be overwritten? That's a good point, since I believe Trados simply replaced the first one...
That behavior is a bit questionable when you have a memory with no duplicate source/different target TUs, a string resource file, for example, and you need to import a memory with updated terms: you'll end up with at least two segments, the original and the updated one.
But you're not 'left alone' on that: you simply do not import the new memory, but rather use it as the 'main memory' with penalty zero in the memories to be used in the translation (you can use as many as you wish), and applies a penalty to your 'unupdated memo', for example "2". When you translate, Studio will use the 100% matches of the new memo, because the old memo will have only 98% matches (100-2).
If you set Studio to "update" both memories during the translation, you'll end up with the 'old' updated too, using the updated terms of the new one. Besides, you can go to the "Memories" pane and 'batch-edit' the old memo with the terms you've just saw in the Editor.
For each client and/or subject I keep one "base memory", maintained by me, verified etc.: when the client sends a job with an updated memory (i.e. "software2009" to "software2010"), I use it as 0% penalty and include the "base memory" with a 2% or 4% penalty (just to see the TUs really penalized on TM pane). In the end I have my base memory updated just on what I really reviewed, not TUs that could have been translated by someone else (the "new" one), and keep the new one just in case of further jobs with the same memory.
It works great for me.
Rgds
Edric
[Edited at 2009-11-22 13:45 GMT]
Thanks Edric for sharing your idea, this is a good solution in the meanwhile and it works great for me as well.
Hopefully those guys from Trados will pay attention to our request, as this is a very important feature to be included in coming updates.
Regards,
Nicolás
Subject:
Comment:
The contents of this post will automatically be included in the ticket generated. Please add any additional comments or explanation (optional)
The contents of this post will automatically be included in the ticket generated. Please add any additional comments or explanation (optional)
Walter Blaser Switzerland Local time: 07:59 Member (2009) French to German + ...
Studio can find duplicate TUs
Oct 21, 2010
Andrei Vybornov wrote:
....
MultiTerm allows to search for duplicates. I wonder why there is no such thing in Studio for TM maintenance. Could save the day...
It can, Andrei. If you are in the view 'Translation Memories' and have a TM open, you can do a search on the TM according to various criteria. In the drop-down field 'Search type', you can select 'Search in potential duplicates only'. This very useful function will return all duplicates, one per page, so you can have a look at them and delete the unwanted TUs. Click on 'Next Page' or hit ALT-right arrow to show the next duplicate and so on.
Walter
Subject:
Comment:
The contents of this post will automatically be included in the ticket generated. Please add any additional comments or explanation (optional)
The translation workplace 
Translation news related to SDL Trados