Looking for a workaround to calculate the really work done
Thread poster: Fernando Toledo

Fernando Toledo  Identity Verified
Germany
Local time: 04:21
German to Spanish
Jun 5, 2005

Everybody knows the problem. A client send you a project based on a TM analysis. Everything seems to be fine. 80 % of the job are 100% matches. So, the invoice will be not so high , but, you think, it will not take too long.
To my surprise almost 50% of the "greens" segments must be changed (bad alignements, big mistakes, not completed sentences from previous translations, etc)

Is there a way to analyze the real made job?
If not with Trados, maybe with another application?

Someone knows if Olifant can do this?
http://www.translate.com/technology/tools/Olifant.html

Thks

Rgds

Toledo



[Edited at 2005-06-05 20:24]


Direct link Reply with quote
 
Aleksandr Okunev
Local time: 05:21
English to Russian
Impossible Jun 5, 2005

Toledo wrote: Is there a way to analyze the real made job?
If not with Trados, maybe with another application?


You want too much from a PC. The machine compares the stats of the segments, that's it, be it nice verse or gibberosh.

I was in the same situation last month when a client tried to give me a machine translated MT with standard fuzzy discount scheme. We renegotiated. Try to supply a couple of examples to you customer, showing bad quality of the 100% matches. If nothing changes, tell him you won't touch the 100% matches or else, abandon.

AFAIK, a TM is a Q-u-a-l-i-t-y C-h-e-c-k-e-d database in normal houses.

HTH
Aleksandr


Direct link Reply with quote
 

Harry Bornemann  Identity Verified
Mexico
English to German
+ ...
Not with Olifant Jun 5, 2005

Your description reminds me to the mobile manual I translated a week ago - and also to the printed manual delivered with my new mobile..

In my case the .txt file with the exported TM had 3.5 MB, hence it would probably take more than a month to correct it (using Olifant, for example).

But you can quickly make the relevant segments visible by pretranslating the file(s) using Workbench/Tools/Translate.

After viewing some of the resulting pages you might change your quote, if you did not give a binding one, yet.


Direct link Reply with quote
 
pep
Local time: 04:21
English to Spanish
Maybe a process with ApSIC Comparator could help Jun 5, 2005

I pictured the following process:

1. Clean-up your final files against the initial memory BUT with the option that does not allow changes to the existing 100% matches. This way the memory will only import new matches and fuzzy macthes, but not 100% matches.

2. Pretranslate the files you received with this memory (everything should be 100% matches).

3. Use ApSIC Comparator to compare your files against those files generated in step 2. If everythigh worked well, the differences should be precisely the 100% matches that you modified. When you run the comparison, define your final files as the "reviewed" files and the pretranslated files as the "unreviewed". In addition to the differences you will also get the number of segments that you had to modify.

Note: This process does not report the work you did on fuzzy matches if the fuzzy matches were of low quality.

I hope this helps.

Pep.


Direct link Reply with quote
 

Heinrich Pesch  Identity Verified
Finland
Local time: 05:21
Member (2003)
Finnish to German
+ ...
You can compare the TM Jun 6, 2005

The segments changed by you would be listed at the end of the TM. You can export them into txt and convert into a table, then count the column which contains the targets.
But if the customer does not pay for correcting the wrong segments I would not touch them.

Regards

Heinrich


Direct link Reply with quote
 

Fernando Toledo  Identity Verified
Germany
Local time: 04:21
German to Spanish
TOPIC STARTER
After years using it Jun 6, 2005

Still having problems to understand the logic on this application.

pep wrote:

1. Clean-up your final files against the initial memory BUT with the option that does not allow changes to the existing 100% matches. This way the memory will only import new matches and fuzzy macthes, but not 100% matche





I am trying to follow your indication, but I do not know how to clean it without changing the 100% matches


there are 4 options and I can not "see" the logic.

Don't update
Update TM
Don't clean -up
Update document




Which is the one I need?
I tried with "Don't update" but the new TUs are not implemented






Thks for your help/time


Toledo

[Edited at 2005-06-07 09:47]


Direct link Reply with quote
 
pep
Local time: 04:21
English to Spanish
Export and then import with Leave unchanged Jun 7, 2005

I was pretty confusing in my explanation, sorry about that.

Let me rewind a little bit:

What you should do is to clean normally in your "gold" memory, then export it to .txt and import the newly created memory into the "garbage" original one with the import option named "Leave unchanged".

I apologize for my lack of precision. I came with the idea becase I had (and have) one client for whom I spent years modifying wrong segments that came from the memory and my client managed to never import my corrections, so I thought the process he followed to avoid importing my corrections was more straight-forward.

I hope this does the trick for you now.

Regards,

Pep.


Toledo wrote:

Still having problems to understand the logic on this application.

I am trying to follow your indication, but I do not know how to clean it without changing the 100% matches

there are 4 options and I can not "see" the logic.

Don't update
Update TM
Don't clean -up
Update document




Which is the one I need?
I tried with "Don't update" but the new TUs are not implemented




pep wrote:

1. Clean-up your final files against the initial memory BUT with the option that does not allow changes to the existing 100% matches. This way the memory will only import new matches and fuzzy macthes, but not 100% matche





Thks for your help/time


Toledo


Direct link Reply with quote
 
pep
Local time: 04:21
English to Spanish
And there is still another variant Jun 7, 2005

I came up with yet another variant much simpler, to get the two set of memories.

1. Once you have the "gold" memory make a second copy and name it "garbage". For now they are the same, but for just a little while.

2. With the original memory (i.e. the one missing new and fuzzy segments), pretranslate the set of files (tools->Translate), with the following settings:

[ 100 ] % or higer match level
Leave unchecked "Segment unknown sentences".

This will create pretranslated files with only original 100% matches. Now you want to clean those matches (Tool->Clean Up->Update TM) into the "garbage" memory that you copied on step 1.

The garbage memory contains your new and fuzzy an their 100% matches.

Now you should be ready to pretranslate again the files with the garbage memory and compare with ApSIC Comparator.

I hope this helps,

Pep.


Direct link Reply with quote
 


To report site rules violations or get help, contact a site moderator:


You can also contact site staff by submitting a support request »

Looking for a workaround to calculate the really work done

Advanced search







PDF Translation - the Easy Way
TransPDF converts your PDFs to XLIFF ready for professional translation.

TransPDF converts your PDFs to XLIFF ready for professional translation. It also puts your translations back into the PDF to make new PDFs. Quicker and more accurate than hand-editing PDF. Includes free use of Infix PDF Editor with your translated PDFs.

More info »
Across v6.3
Translation Toolkit and Sales Potential under One Roof

Apart from features that enable you to translate more efficiently, the new Across Translator Edition v6.3 comprises your crossMarket membership. The new online network for Across users assists you in exploring new sales potential and generating revenue.

More info »



Forums
  • All of ProZ.com
  • Term search
  • Jobs
  • Forums
  • Multiple search