Mobile menu

Confusing Analysis Results (Use TM from previous analysis)
Thread poster: Stefan Gentz

Stefan Gentz
Local time: 15:07
English to German
+ ...
Mar 3, 2006

[SDL TRADOS Workbench 7.1.0.719]

Hello,

I have a problem understanding TRADOS analysis results.

Situation:
There are two manuals, that have to be translated. There is no TM existing. The client agrees, to translate the manuals one by one and not parallel to take advantage of existing translation from the first manual in the second manual.

I tried the following:
First, I did an analysis of the first manuals's TTX files based on an empty TM. I deleted the ttx files ("Clear") in the "Analys Files" dialog. Then I drag & dropped the TTX files from the second manual into the analysis dialog and activated the "Use TM from previous analysis" checkbox. I ran the analysis again and – as expected – got a result with a lot of fuzzy matches.

Now I did the following for testing purposes:
With the empty TM I "translated" the TTX files from the first manual. Actually I only simulated a translation by using the "Segment unknown sentences". The result are TTX files with source and target segments with the source text copied into the target segment. Then I cleaned the TTX files and got a TM with the DE-DE - DE-DE segemts.

And now the surprise:
With this TM containing the segments from the first manual I did again an analysis of the first manual. However the results are now completely different from the first analysis with the "Use TM from previous analysis" results.
Actually you would expect that the result is 100% identical. However, the results are very different: E.g. the "Use TM from previous analysis" results gives me 287 words for the 95% - 99% matches. The analysis based on the TM created with the "Segment unknown sentences" and clean up gives me only 2 words for the 95% - 99% matches.

My conclusion is, that there is a fierce bug in the way TRADOS analysis files.
Either the "Use TM from previous analysis" feature works wrong or the "normal" analysis feature has a problem.

Also I'm wondering about the following analysis approach: If I have not only two but three or more manuals to analyse, does the "Use TM from previous analysis" feature only count the third manual based on the analysis of the second manual? Or does this also include possible matches based on manual one + manual 2?

Any ideas?

Best,
*Stefan.

[Subject edited by staff or moderator 2006-03-03 09:57]


Direct link Reply with quote
 

PCovs
Denmark
Local time: 15:07
Member (2003)
English to Danish
+ ...
I am confused Mar 3, 2006

Stefan Gentz wrote:

I tried the following:
First, I did an analysis of the first manuals's TTX files based on an empty TM. I deleted the ttx files ("Clear") in the "Analys Files" dialog. Then I drag & dropped the TTX files from the second manual into the analysis dialog and activated the "Use TM from previous analysis" checkbox. I ran the analysis again and – as expected – got a result with a lot of fuzzy matches.

---

And now the surprise:
With this TM containing the segments from the first manual I did again an analysis of the first manual.

Best,
*Stefan.

[Subject edited by staff or moderator 2006-03-03 09:57]


When I read this, I conclude that you have first analysed the first manual, then second manual getting this or that result, then you "translate" the first manual, and then you analyse the FIRST manual again? I suppose you must have meant the SECOND manual, right?


Direct link Reply with quote
 

Stefan Gentz
Local time: 15:07
English to German
+ ...
TOPIC STARTER
Sorry ... Mar 3, 2006

... you are correct. It should read:
With this TM containing the segments from the first manual I did again an analysis of the second manual.


Best,
*Stefan.


Direct link Reply with quote
 

PCovs
Denmark
Local time: 15:07
Member (2003)
English to Danish
+ ...
Anyway, I don't understand you. Mar 3, 2006

Why would you want to analyse the two manuals this way?

If you analyse all the files in both manuals at the same time, you will have all the repetitions etc. all-in-one, but perhaps you need to have separate analyses for your customer?


Direct link Reply with quote
 

Stefan Gentz
Local time: 15:07
English to German
+ ...
TOPIC STARTER
Analysis of all files in one batch vs. "Use TM from previous analysis" Mar 3, 2006

PCovs wrote:
If you analyse all the files in both manuals at the same time, you will have all the repetitions etc. all-in-one,
quote]

That's exactly the point. If I analyse all files from all manuals in one batch, I get only the repetitions. But that doesn't tell me (or the client) who much he will save, if he first translates manual 1 and after this manual 2. If you analyse both manuals with an empty TM, you don't get fuzzy matches.
The topic is about simulating a growing TM in advance.

Finally it's very simple:
Option 1:
I analyse manual 1 and 2 in one batch with the empty TM and I get analysis result x. The resulting quote can take advantage of the repetitions, but not of fuzzy matches for manual 2.

Option 2:
If I translate manual 1, close this job and do an analysis of manual 2 with the TM (that now contains the complete translation of manual 1), I will get a very different analysis for manual 2 compared with the inital analysis of option 1.

The task is to analysis manual 2 in advance, simulating an existing translation of manual 1. To do this, there is the "Use TM from previous analysis" in TRADOS Workbench.
And this is where my problem starts! The analysis results for manual 2 with "Use TM from previous anaylsis of manual 1" based on an empty TM are very different from an analysis of manual 2 based on an TM containing the translation of manual 1.

I hope the problem is more clear now?

Best,
*Stefan.


Direct link Reply with quote
 

Heinrich Pesch  Identity Verified
Finland
Local time: 16:07
Member (2003)
Finnish to German
+ ...
Confusing indeed Mar 3, 2006

No matter what way you translate the two docs, the number of matches is the same.

Direct link Reply with quote
 

Jabberwock  Identity Verified
Poland
Local time: 15:07
Member (2004)
English to Polish
Find the difference... Mar 3, 2006

I understand the problem, in fact I have often used that technique to determine the amount of work which a file needs.

You might try to output the "unknown segments" with different settings from the two analysis options. That way you could compare them and see which segments are treated differently...

Unfortunately I do not have much time now, but I will run some experiments of my own...


Direct link Reply with quote
 
tyshelley
United States
Local time: 08:07
Spanish to English
I know this is an old string... but resurrecting it might help me out. Sep 16, 2011

I know this is an old string... but resurrecting it might help me out.

The analysis method referred to in the Trados manual as "Use TM from previous analysis" method has turned out to be a very helpful method, depending on the type of project. As described, it's best for deciding how to quote when translating multiple yet similar files. The ideal method is to translate the file first that contains the bulk of what is in the rest of the files and then reuse translations. Simply analyzing all files together won't yield the same results and in the end too much will be paid out for translation, or too much will be charged by the translator.

Admittedly, quoting for a client with this method means you have to get the translators on board with you or your profit margin will suffer (coming from an agencies point of view). Most translators I've worked with, as intelligent as they may be, tend to struggle with the logic of an analysis method that will lead them to getting paid less, understandably so... they still have to look at the text and make decisions even where the text is the same.

The discrepancy you're seeing in the results does seem to be bug related... or possibly has to do with the penalty settings. Something to look into to.

Now, to further complicate the matter: I'm trying to figure out how to perform the "Use TM from previous analysis" method in SDL Studio 2009. We've recently upgraded in the office and are trying to get the most out of having it in-house. I've figured out how to perform a "traditional" analysis w/out too much trouble. But I'm getting no where with using a TM from a previous analysis. Any help is greatly appreciated.


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 »

Confusing Analysis Results (Use TM from previous analysis)

Advanced search


Translation news related to SDL Trados





Anycount & Translation Office 3000
Translation Office 3000

Translation Office 3000 is an advanced accounting tool for freelance translators and small agencies. TO3000 easily and seamlessly integrates with the business life of professional freelance translators.

More info »
CafeTran Espresso
You've never met a CAT tool this clever!

Translate faster & easier, using a sophisticated CAT tool built by a translator / developer. Accept jobs from clients who use SDL Trados, MemoQ, Wordfast & major CAT tools. Download and start using CafeTran Espresso -- for free

More info »



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