Use Trados to Quote Thread poster: Bupalos
|
Hi Everyone, great looking community here... I wonder if anyone can help me with the issue of the best way to use trados to analyse files prior to translating for quoting purposes. When I do an analysis on the source files, all it seems to do is give back exact matches- practically nothing...but the documents are full of very similar sentences. Can't it easily give the fuzzy match breakdown too? The client wants to see the analysis beforehand. | | |
No TM no fuzzies! | Apr 26, 2005 |
You get fuzzies only if there are similar TUs in your TM, with a empty TM you only get Repetitions. Rgds Toledo | | |
Hmmm, that seems pretty silly... | Apr 26, 2005 |
What you say corresponds with what I see--and I guess that's how trados is-- but that seems to make no sense. I would actually have to start the job before I could make any comments about how much similarity there is. It should be able to compare segments to one another regardless of whether they have been translated before. Can any of the other CAT tools do this? | | |
Ralf Lemster Germany Local time: 06:28 English to German + ... How many documents do you need to analyse? | Apr 26, 2005 |
Hi Bupalos, Toledo is right - fuzzy matches are analysed against a TM; only repetitions are shown on top of this. That's as designed, actually, as you will need to work these segments. If the number of documents is limited, or if there is one document (or a limited number of docs) you could use as 'master document(s)', you could try this: - Analyse the master doc(s). Don't close the dialogue. - Check Use TM from previous analysis. - Remove the a... See more Hi Bupalos, Toledo is right - fuzzy matches are analysed against a TM; only repetitions are shown on top of this. That's as designed, actually, as you will need to work these segments. If the number of documents is limited, or if there is one document (or a limited number of docs) you could use as 'master document(s)', you could try this: - Analyse the master doc(s). Don't close the dialogue. - Check Use TM from previous analysis. - Remove the analysed file(s), and add the remaining files. - Run another analysis. The results will give you overlaps between the first and second groups of documents. I would strongly suggest to test various scenarios before using this to quote. Also, you may want to ask yourself whether you really need to 'give away' this efficiency (which is not revealed by a 'normal' analysis). Best regards, Ralf ▲ Collapse | |
|
|
You can do this with SDLX | Apr 26, 2005 |
Hi! Can any of the other CAT tools do this? You can do this with SDLX - only with the Professional version, though, because you need the Analyse function which is not included in the Standard version. You can download a fully functional 30-day trial version from the SDL web site. You will need to use the Analyse function of the software and make sure that the Simulate Auto-propagation box is checked on the Options tab. Hope this helps. Regards, Claudia | | |
Thanks guys! | Apr 26, 2005 |
"Also, you may want to ask yourself whether you really need to 'give away' this efficiency (which is not revealed by a 'normal' analysis)." Unfortunately, the client is requiring vendors to show what efficiency they will be delivering upfront as part of the bidding process. Good suggestion on the "serial" approach. I hate to add extra work for myself to reduce the amount of the final job, but it seems this is our lot nowadays. I at least if I do it that way, I will inc... See more "Also, you may want to ask yourself whether you really need to 'give away' this efficiency (which is not revealed by a 'normal' analysis)." Unfortunately, the client is requiring vendors to show what efficiency they will be delivering upfront as part of the bidding process. Good suggestion on the "serial" approach. I hate to add extra work for myself to reduce the amount of the final job, but it seems this is our lot nowadays. I at least if I do it that way, I will include the extra work to give this up front. Claudia, thanks so much for that suggestion. I will try that first. You guys are absolutely great. I'm looking forward to participating here more as I gradually lose my status of village idiot. ▲ Collapse | | |
SDLX approach didn't seem to work | Apr 26, 2005 |
I downloaded the SDLX, converted a couple test word files to rtf, identified the file set as the "new files", marked the auto-propagation option and the cross-file option, and ran the analysis. The files have four sentences each, with one an exact duplicate and three just barely changed. Here were the options: New Files: Selected : 2 Old Files: Selected : 0 Translation Memories: No translation memories selected. ... See more I downloaded the SDLX, converted a couple test word files to rtf, identified the file set as the "new files", marked the auto-propagation option and the cross-file option, and ran the analysis. The files have four sentences each, with one an exact duplicate and three just barely changed. Here were the options: New Files: Selected : 2 Old Files: Selected : 0 Translation Memories: No translation memories selected. Translation Memory Options: - Minimum Match Value: n/a - Context Mismatch Penalty: 1% - Filter Mismatch Penalty: 1% - Using Target Language: n/a Autopropagation Options: - Enabled: Yes - Minimum Match Value: 75% Cross File Repetitions: - Enabled: Yes Automatch Options: - Enabled: Yes - Dates: Yes - Numbers: Yes - Acronyms: Yes General Options: - Word Count Algorithm: Standard - Hide Number Segments: No - Segmentation Rules: Default Starting analysis process.... It gave me back 0 replications, 0 matches, 0 on all the fuzzies. Makes me feel like a 0. ▲ Collapse | | |
You need to convert the files to otf files first | Apr 27, 2005 |
It gave me back 0 replications, 0 matches, 0 on all the fuzzies.
Ok, I think I know why this happened. You need to convert the files to otf format in order to be able to analyse them. Sorry, I forgot to mention this yesterday, it was just too late for my brain to work properly. Here's what you do: You convert the files into itd translation files using the Project Wizard. (This process is quite straight-forward so I won't go into details, just give me a shout if you come across any problems.) The important thing is to deselect the Remove tag file when itd is generated check box in the last but one dialog. This way the programm will save the otf files that it creates during conversion (they will be saved in the folder of your original files). Use these otf files for the analysis, do everything the way you did it with the rtf files (your procedure was fine) and you should get a log file with replications and fuzzies. I just tried it and it worked fine. HTH Regards, Claudia | |
|
|
Everything seemed to work... | Apr 27, 2005 |
but I still get zeros across the board. As I say, I set up the test files and I know the segments should range between 80-95% fuzzy match. In the report file, it puts a (TM) after each category, thus-- Filename 100% (TM) 95% - 99% (TM) 85% - 94% (TM) 75% - 84% (TM) Repetitions Untranslated Total Words C:\Trados_projects\test\source\This is a test of the new_doc.otf 0 0 0 0 0 79 79 C:\Trados_projects\test\source\This is a test_doc.otf 0 0 0 0 0 76 76 To... See more but I still get zeros across the board. As I say, I set up the test files and I know the segments should range between 80-95% fuzzy match. In the report file, it puts a (TM) after each category, thus-- Filename 100% (TM) 95% - 99% (TM) 85% - 94% (TM) 75% - 84% (TM) Repetitions Untranslated Total Words C:\Trados_projects\test\source\This is a test of the new_doc.otf 0 0 0 0 0 79 79 C:\Trados_projects\test\source\This is a test_doc.otf 0 0 0 0 0 76 76 Totals 0 0 0 0 0 155 155 The (TM) makes me think it is indicating this is matches against the memory. When it is asking me what memory to use for the analysis, I am just leaving that blank. Maybe that is the problem...do I need to have a memory stipulated or something? Thought I had it there...was almost ready to break out the wine. ▲ Collapse | | |
One more thing you need to do... | Apr 27, 2005 |
Thought I had it there...was almost ready to break out the wine. We're getting there - there is one more thing you need to do... If you start the Analyse function, make sure that when you have entered a log file name on the Reporting tab, you click on the Field Order button. You will need to select all the match values that contain an AP (for auto-propagation) in brackets to be included in the report. I think you also have to click on the Export button once but I'm not 100% sure about this. Just try it, it definitely won't do any harm Regards, Claudia | | |
Claudia, I wood sing yurr praises | Apr 27, 2005 |
but I am too druunk with happeeNess to tYpe! When I sober up I'll write a paen to Claudia, the Goddess of SDLX. For now, A VERY BIG THANK YOU! (Can you tell it worked?) | | |
I'm glad it worked! | Apr 27, 2005 |
Just one last thing: you won't need to do this Field Order and Export thing every time you analyse files, the settings will be saved in SDLX and in the future auto-propagated matches will automatically appear in your log files. - That's why I didn't even remember you had to do this. From now on, all you need to do is create otf files from your source files and analyse them. Regards, Claudia | | |