Stran v tématu: [1 2] > | WF Pro: what are "fuzzy reps"? Autor vlákna: pcs_MCIL
|
My long term client just told me about "fuzzy reps" explaining that are a new class of repetitions in the latest versions of WF Pro. He explained me they are repetitions within the same document, and provided its analysis showing: - 100% matches and repetitions (from the existing TM) - fuzzy reps - fuzzy - no match I tried to google this term, and run an analysis in WF Pro 3.1.4 and 3.1.5 but they don't appear. Also, my client asked me if I enabled fuzzy reps. I am kinda clueless here. Is this the same thing as self-propagation? ▲ Collapse | | | Rolf Kern Švýcarsko Local time: 03:42 angličtina -> němčina + ... In memoriam I would say: | Mar 28, 2013 |
fuzzy reps = more or less exact repetitions fuzzy = similar repetitions However, I do not use any CAT tools and am therefore not familiar with these funny distictions. Rolf
[Bearbeitet am 2013-03-28 20:50 GMT] | | | | Heinrich Pesch Finsko Local time: 04:42 Člen (2003) finština -> němčina + ... A new trick to lower rates? | Mar 29, 2013 |
Of course if the document contains a lot of these fuzzy reps translating it becomes easier. But these schemes do not help us freelancers, instead they allow customers to insist on lower rates. So if a tool like WFP does show this new feature they try to get more corporate users for WFP. We should try to block such attempts. | |
|
|
neilmac Španělsko Local time: 03:42 španělština -> angličtina + ...
Rolf Kern wrote: fuzzy reps = more or less exact repetitions fuzzy = similar repetitions Rolf
[Bearbeitet am 2013-03-28 20:50 GMT] My understanding of the terms is the same as Rolf's*. I don't know whether it has a more specific meaning internally in CAT programs, but I don't think I need to. Caveat: the concept is applied by agencies or similar clients who want to whittle away at your potential earnings for their own benefit. I tend to avoid this kind of working situation and prefer to work at an average-to-low rate but without any penalties or reductions for repetitions, fuzzy or otherwise. I do however, allow my clients a discretionary discount when I feel there are enough repetitions in the text to merit it, but it's my decision, not something I have to "like or lump". *Cf: "fuzzy logic", a field I have managed to translate and revise texts in without really understanding much about. | | | neilmac Španělsko Local time: 03:42 španělština -> angličtina + ... No new trick | Mar 29, 2013 |
Heinrich Pesch wrote: Of course if the document contains a lot of these fuzzy reps translating it becomes easier. But these schemes do not help us freelancers, instead they allow customers to insist on lower rates. So if a tool like WFP does show this new feature they try to get more corporate users for WFP. We should try to block such attempts. My point exactly - and the notion has been around since I got into translating 15 years ago. | | |
Rolf Kern wrote: fuzzy reps = more or less exact repetitions fuzzy = similar repetitions However, I do not use any CAT tools and am therefore not familiar with these funny distictions. Rolf I'm afraid this is not the case, Rolf. - Fuzzy matches are less than 100% matches between a document and your current TM - Fuzzy reps (or internal fuzzies, as Dominique writes) are fuzzy matches that only become fuzzies as you translate and build up your TM, i.e. fuzzy matches within a document/project. | | | pcs_MCIL angličtina -> italština + ... AUTOR TÉMATU I thought it was already so | Mar 29, 2013 |
Thank you very much for your explanation, now it is all clear. It seems odd that this agency didn't use this trick before. For this same project, I was initially told it had a certain amount of words, then when the PO came these fuzzy reps were introduced and accounted for 1/4 of the job, so it looked quite unfair. To me, a fuzzy is a fuzzy, no matter if it comes from the previous TM or from this same document. I... See more Thank you very much for your explanation, now it is all clear. It seems odd that this agency didn't use this trick before. For this same project, I was initially told it had a certain amount of words, then when the PO came these fuzzy reps were introduced and accounted for 1/4 of the job, so it looked quite unfair. To me, a fuzzy is a fuzzy, no matter if it comes from the previous TM or from this same document. It requires the same effort for the translator. Of course, it is important to understand the fuzzy logic (99%, 75%, 50% or else). However I thought the analysis in WF Pro was already accounting for internal fuzzies by default, since in the past I analyzed a file(s) against an empty TM and it returned several fuzzies, that could come only from that same file(s) being the TM empty. @Dominique, I watched your video and it was very helpful. @all the others: I agree that agencies hold to fuzzies as if they were the ultimate source to lower their cost (= translator's fee) and are pulling all sort of tricks to increase the number of fuzzies by pre-translating the document against clusmy TMs. For example, in this TM there were several segments from an alignment that resulted in wrong source-target entries. At a certain point this was so obvious... there was a list of the countries of the world and most of them were mistranslated and counted as fuzzies. If Australia is pre-translated as Bolivia, then it's no fuzzy at all! I could argue with the client and he would ask me to quantifies these mis-fuzzies, that would take me away more time than it's worth because it would a human-job, subjective, etc. I am getting tired of these tricks. CAT do help translators a lot, but using a computer it's not all there is to translation!
[Edited at 2013-03-29 10:06 GMT] ▲ Collapse | |
|
|
Instead of Crying! | Mar 29, 2013 |
Dear Girino, Heinrich and All, I mentioned before that this feature should NOT be used by agencies for lowering rates paid to translators. It is only meant for helping organise resources and schedules. In my humble opinion, translators (not agencies) are who spoil the translation market by accepting such unfair procedures and rates. Instead of just crying and accepting the unfair rate, one should publicly name such unprofessional agencies in all sorts of "Fame and Shame... See more Dear Girino, Heinrich and All, I mentioned before that this feature should NOT be used by agencies for lowering rates paid to translators. It is only meant for helping organise resources and schedules. In my humble opinion, translators (not agencies) are who spoil the translation market by accepting such unfair procedures and rates. Instead of just crying and accepting the unfair rate, one should publicly name such unprofessional agencies in all sorts of "Fame and Shame" sections and ask Associations for warning translators about them - they should be announced as "banned" agencies. Kind regards, Yasmin Girino wrote: For this same project, I was initially told it had a certain amount of words, then when the PO came these fuzzy reps were introduced and accounted for 1/4 of the job, so it looked quite unfair. ▲ Collapse | | | pcs_MCIL angličtina -> italština + ... AUTOR TÉMATU I run my analysis again and... | Mar 29, 2013 |
Yasmin, nobody is blaming the tool. Wordfast is great, 98% of my jobs is on WF Pro with this agency, so I can't really complain too much. I run again my analysis including internal fuzzy matches but the new word count is ten times the former. The agency's word count was 15k no match and 9k fuzzies (5k internal + 4k external). My word count was 20k new and 4k fuzzies (external only) My new word count with internal fuzzies is 50k no match and 180k fuzzies, whi... See more Yasmin, nobody is blaming the tool. Wordfast is great, 98% of my jobs is on WF Pro with this agency, so I can't really complain too much. I run again my analysis including internal fuzzy matches but the new word count is ten times the former. The agency's word count was 15k no match and 9k fuzzies (5k internal + 4k external). My word count was 20k new and 4k fuzzies (external only) My new word count with internal fuzzies is 50k no match and 180k fuzzies, which is clearly impossible. Yasmin, if I send you the file can you please clarify this for me once and for all? I am using WF Pro 3.1.5. ▲ Collapse | | | Samuel Murray Nizozemsko Local time: 03:42 Člen (2006) angličtina -> afrikánština + ... No, it is a fairly new thing | Mar 29, 2013 |
Girino wrote: However I thought the analysis in WF Pro was already accounting for internal fuzzies by default, since in the past I analyzed a file(s) against an empty TM and it returned several fuzzies, that could come only from that same file(s) being the TM empty. In my WFP, the option to count internal fuzzies is deselected by default, but perhaps it was selected in your case. Anyway, not all CAT tools can count internal fuzzies, and I think in earlier days internal fuzzies were not included in the "fuzzy" totals. This means that if you did an analysis against an empty TM, there would be zero fuzzies. I have said this before, but I'll say it again, namely: I think requesting discounts for internal fuzzies is more fair than requesting discounts for external fuzzies, because external fuzzies often come from other translators whose translation quality you can't be certain of, whereas internal fuzzies you yourself have full control over from the start. If you are a cautious translator, internal fuzzies should take you less time to fix than external fuzzies, unless the TM you got is very good. It is said by some that 100% matches should be free unless the client wants the translator to actually them. In the same vein, external fuzzies should only be discounted if the client is happy that only the barest of necessary changes are made to them and errors that existed in the translation memory need not be fixed by the translator. | | | pcs_MCIL angličtina -> italština + ... AUTOR TÉMATU Unselected by default | Mar 29, 2013 |
Samuel, I agree with everything you said. I am sure my fuzzies are a "real" fuzzies, whereas a fuzzies from a clusmy TM (like the one I was given) are useless. I also believe that short segments should not concur in the fuzzy count, since the shorter the segment, the less it is likely to be a "real" fuzzy. Also, when translating into Latin-derived languages, there are so many variants (gender, number, formal/informal register) that short fuzzy segments are more no-match ... See more Samuel, I agree with everything you said. I am sure my fuzzies are a "real" fuzzies, whereas a fuzzies from a clusmy TM (like the one I was given) are useless. I also believe that short segments should not concur in the fuzzy count, since the shorter the segment, the less it is likely to be a "real" fuzzy. Also, when translating into Latin-derived languages, there are so many variants (gender, number, formal/informal register) that short fuzzy segments are more no-match segments. The internal fuzzy count is unselected by default in WF Pro, but I think it has been there for at least one year. I didn't notice it before because the difference between my analysis and the agency analysis has not been so striking as with a 25k words file like this one. ▲ Collapse | |
|
|
Samuel Murray Nizozemsko Local time: 03:42 Člen (2006) angličtina -> afrikánština + ... Post-translation analysis | Mar 29, 2013 |
Girino wrote: Also, when translating into Latin-derived languages, there are so many variants (gender, number, formal/informal register) that short fuzzy segments are more no-match segments. I believe it would be good business practice to perform a post-translation analysis to see if what the fuzzy count is if it is based on the target text, not the source text. In other words, at the end of the translation, reverse the languages of the original TM and then run an analysis of the translation against that TM (including internal fuzzies). Of course, some translators might think to cheat such a system by making unnecessary changes to the fuzzy matches, but that is the same as saying that translators who charge by target word count will try to cheat the client by using more words in the translation. | | |
Dear Samuel, Samuel Murray wrote: I have said this before, but I'll say it again, namely: I think requesting discounts for internal fuzzies is more fair than requesting discounts for external fuzzies, because external fuzzies often come from other translators whose translation quality you can't be certain of, whereas internal fuzzies you yourself have full control over from the start. If you are a cautious translator, internal fuzzies should take you less time to fix than external fuzzies, unless the TM you got is very good. This makes sense to me. However, such client will rather ask for discounts for both external and internal fuzzies; will this be fair in such case? Another case: I heard about clients considering 40% matches as fuzzies that require discounts. Does this really make any sense? Well, I believe those matters should be rather discussed by Associations to have clear standards; otherwise, it is up to everyone to evaluate matters based on their own circumstances, which is justifiable, but not perfect. Kind regards, Yasmin | | | With pleasure | Mar 29, 2013 |
Dear Girino, Girino wrote: Yasmin, if I send you the file can you please clarify this for me once and for all? I am using WF Pro 3.1.5. With pleasure! I hope I can help. Just please make sure I will be able to understand your precise question. Thanks! Kind regards, Yasmin | | | Stran v tématu: [1 2] > | To report site rules violations or get help, contact a site moderator: You can also contact site staff by submitting a support request » WF Pro: what are "fuzzy reps"? 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 Trados, MemoQ, Wordfast & major CAT tools.
Download and start using CafeTran Espresso -- for free
Buy now! » |
| Trados Studio 2022 Freelance | The leading translation software used by over 270,000 translators.
Designed with your feedback in mind, Trados Studio 2022 delivers an unrivalled, powerful desktop
and cloud solution, empowering you to work in the most efficient and cost-effective way.
More info » |
|
| | | | X Sign in to your ProZ.com account... | | | | | |