Pages in topic:   [1 2 3] >
What are "fuzzy matches" for?
Thread poster: LilianNekipelov

LilianNekipelov  Identity Verified
United States
Local time: 06:11
Russian to English
+ ...
Aug 13, 2013

I have been wondering for a long time what the so called "fuzzy matches" are for? What sense do they have -- if something is imprecise, it is basically wrong. I personally think they are totally useless in translation, and they can only confuse the translator more by suggesting only seemingly similar phrases.

Another thing, why would experienced, professional translators be expected to use some else's TMs -- this is what some agencies require. Are we supposed to check the TMs first if they are correct? This may be more of a lexographical task. Why would anyone want to complicate our lives, instead of making them easier, and more enjoyable?


 

Yasutomo Kanazawa  Identity Verified
Local time: 19:11
As the term explains itself, Aug 13, 2013

fuzzy matches are similar matches which are not totally identical.

I understand your frustration, and like every coin has two sides, fuzzy matches also have the pros and cons. Not all fuzzy matches are easy and simple sentences, such as :

"This is a ball-point pen." and "This is a fountain pen.", where the only difference between these two sentences are the word "ball-point" and "fountain". For these kind of sentences with simple structures, the fuzzy parts are easy to identify. But if we have a more complex sentence, it will just confuse the translator like you said.

I guess there is no solution as long as you are using a CAT tool, and what frustrates me more is that they are subject to standardized "CAT tool discounts" where sometimes it is easier to translate the whole sentence from scratch than looking and replacing them with the correct translations.


 

gianfranco  Identity Verified
Brazil
Local time: 07:11
Member (2001)
English to Italian
+ ...
Above 90% fuzzy matches are very often useful Aug 13, 2013

Most modern CAT tools highlight the difference (added words, missing words, punctuaction, etc.) between the segment being processed and the fuzzy match found in memory.

In general, it is not a problem to adapt the match found to the current sentence, but, I'd say, this is true mostly for matches evaluated at least 85% or even 90% and above.
Anything below this threshold, I agree, is of very limited use.

So, I agree with the colleague starting the thread for anything below 85-90%, but I see fuzzy matches as a good technique and resource, to speed up our work and keep consistency, for matches above 90%.

GF


 

LilianNekipelov  Identity Verified
United States
Local time: 06:11
Russian to English
+ ...
TOPIC STARTER
In which way can they be helpful? Aug 13, 2013

Can you give me some examples? If you look at language form this point of view it may turn out to be one, big fuzzy match -- a collection of apparent repetitions.


[Edited at 2013-08-13 14:33 GMT]


 

Maciej W  Identity Verified
Poland
Local time: 12:11
English to Polish
+ ...
Example Aug 13, 2013

The first thing that comes to mind: segments that are identical (or nearly identical) except for a single variable. If you often work with a particular type of texts, these are likely to appear on a daily basis. For example, contracts: "The Employee will receive a fixed salary in the amount of ..." (I think all CAT tools will take care of the number for you, too). Software manuals: "Type XXXX in the Command field and press Enter". Etc. I have done quite a few of both, and have found fuzzy matches, in some cases as "weak" as 70%, to be extremely helpful and time saving.

 

neilmac  Identity Verified
Spain
Local time: 12:11
Spanish to English
+ ...
All the better to screw you with... Aug 13, 2013

I see fuzzy matches as yet another stick agencies can (and often do) use to beat us freelancers over the head with and whittle away at our potential earnings, all in their favour, of course.

As far as I'm concerned, they can stick their fuzzy matches up their... jumpers.

[Edited at 2013-08-13 15:04 GMT]


 

LilianNekipelov  Identity Verified
United States
Local time: 06:11
Russian to English
+ ...
TOPIC STARTER
Doesn't it take less time to just type the phrase Aug 13, 2013

Maciej W wrote:

The first thing that comes to mind: segments that are identical (or nearly identical) except for a single variable. If you often work with a particular type of texts, these are likely to appear on a daily basis. For example, contracts: "The Employee will receive a fixed salary in the amount of ..." (I think all CAT tools will take care of the number for you, too). Software manuals: "Type XXXX in the Command field and press Enter". Etc. I have done quite a few of both, and have found fuzzy matches, in some cases as "weak" as 70%, to be extremely helpful and time saving.


than to figure out what to press, and what can go wrong? I know the ability to repeat certain boilerplates -- identical segments of one paragraph or more is useful in legal translation, but other than that -- I am not sure.


 

Maciej W  Identity Verified
Poland
Local time: 12:11
English to Polish
+ ...
Matter of preference Aug 13, 2013

LilianBNekipelo wrote:

Doesn't it take less time to just type the phrase than to figure out what to press, and what can go wrong?



It might. It might not. It doesn't matter, actually. In my world retyping, let's say, a 300-character phrase you could just insert and edit with a dozen keystrokes makes as much sense as retyping a good, 100% match. At the same time, I accept that you'd rather type the thing and be done with it rather than wonder if you've done everything right. I, for example, tend to avoid typing URLs I've already visited, too - in most cases it's more comfortable to type one, two, or three characters and choose the one I want from the list. You could say I'm just not big on typing, but I'm as far from proselytizing.


As to the agency "discounts" on matches - no translator likes them, but in my experience, short as it may be, their disadvantages are more than balanced out by the advantages of leveraging the matches in work for direct clients.

[Edited at 2013-08-13 15:35 GMT]


 

LilianNekipelov  Identity Verified
United States
Local time: 06:11
Russian to English
+ ...
TOPIC STARTER
Yes, matches can make sense in some types of translation, Aug 13, 2013

like various lists of terms, catalogues, etc. They do not make much sense in many other types of translation. It should also be clearly stated what the company, or a particular program, counts as repetitions.

Another point -- where do these suggestions come from? Who created them? Are they usually accurate? Do they come just form your own TM?


[Edited at 2013-08-13 16:06 GMT]


 

philgoddard
United States
Member (2009)
German to English
+ ...
A very good question, Lilian Aug 13, 2013

I recently asked an agency whether they were passing on fuzzy discounts to their customers, and they said yes. I said, very politely, that I didn't see the point of investing in technology that (in theory, at least) improves your productivity, and then reducing your charges to your customers. After all, we didn't cut our rates when we switched from typewriters to wordprocessors.

[Edited at 2013-08-13 18:08 GMT]


 

Christine Andersen  Identity Verified
Denmark
Local time: 12:11
Member (2003)
Danish to English
+ ...
Set the price up! Aug 13, 2013

I find fuzzy matches useful, but they have to be carefully checked in the new context.
I can agree to a discount for large sections of perfect matches - which are rare in the work I do. But normally I set my rates higher in the first place if agencies want a discount!

After all, I pay for the CAT tool and courses of instruction on how to use it. Worth every penny, but it is still money out of my pocket. I am also picky about which CAT I will use.

I love some and hate some, depending on what features are available or missing, and how well they work.

We are business partners, not employees, so if agencies want our services, they have to go along with the way we provide them.

If that means aligning a text AFTER it has been translated, then so be it. It's not that difficult!


 

Oleg Karnaushenko  Identity Verified
Ukraine
Local time: 13:11
English to Ukrainian
+ ...
Usefullness of fuzzy matches Aug 14, 2013

Hi Lilian,

Just imagine the following situation: you are translating some camera manual for a new camera model that introduced, say "pets group photo" shooting mode. And the section describing this mode is almost identical to the one about group portrait mode, which exists in previous camera models and was translated before. Changing mode name is all but only difference. It would be impractical to translate entire section from scratch and not to leverage previous translation.
Moreover, if you translate it from scratch using your own style, final document will look like quilt, making poor experience for the end user.
Of course, fuzzy matches, and CAT tools in general are not usefull in literary translation, for example, where consistency is an evil rather then goodness

[Відредаговано 2013-08-14 07:12 GMT]


 

tnieminen
Finland
Local time: 13:11
English to Finnish
Good description of language Aug 14, 2013

LilianBNekipelo wrote:

Can you give me some examples? If you look at language form this point of view it may turn out to be one, big fuzzy match -- a collection of apparent repetitions.


[Edited at 2013-08-13 14:33 GMT]


That's what language is like, any utterance must be mostly repetition (lexically or syntactically) to be understandable by others. Uttering a bunch of novel words in a novel syntactic order is the same as speaking in a foreign language.

Personally I've found that the fuzzy discounts expected by agencies correlate fairly well with the actual changes in workload. You have to learn to work with a TM, though. The most important thing is to not fix everything to match your own style of writing. You need to be able to evaluate the suitability of a translation without letting your own stylistic preferences (which usually are just whatever you happened to learn at some critical stage of your career) or perfectionism cloud your judgement.


 

José Henrique Lamensdorf  Identity Verified
Brazil
Local time: 07:11
English to Portuguese
+ ...
The problem is not in the fuzzy matches themselves Aug 14, 2013

LilianBNekipelo wrote:

I have been wondering for a long time what the so called "fuzzy matches" are for? What sense do they have -- if something is imprecise, it is basically wrong. I personally think they are totally useless in translation, and they can only confuse the translator more by suggesting only seemingly similar phrases.


IMHO the problem lies in translation agencies wanting to pay less than no-matches for them.

If we extrapolated this mathematically ad infinitum for a reduction ad absurdum, the final statement (from the agency) would be, As you speak English fluently, we won't pay you anything to translate into English. We'll only pay you for what you translate into, say, Swahili".

My pet analogy is with automatic transmission in a car. When I board a cab, I don't expect to pay less for the same ride because the vehicle has automatic transmission and/or power steering (both cost extra to the owner/operator, not me), which will make the job less tiresome for the driver.

I'd tell each PM who demanded a hefty discount on fuzzy matches to take a cab with automatic transmission and try negotiating a discount with the cabbie on that. That PM would end up taking a bus.

While I give repeated segments for free* on jobs above 5,000 words, I never give any discounts on fuzzy matches; customers demanding them simply don't make it to my clients list.

*The fact is that in these cases I won't translating anything; the computer is copying and pasting automatically. It's similar to not charging extra for hyphenation.

The only reason why discounts on fuzzy matches exist is that one developer has been consistently boasting the savings they offer to sell more, greedy translation agencies believe it, AND naive translators accept it. If any of these three conditions failed to prevail, such discounts would not exist.


 

LilianNekipelov  Identity Verified
United States
Local time: 06:11
Russian to English
+ ...
TOPIC STARTER
Well, I have never seen too many repetitions Aug 14, 2013

in my types of translation, except for some catalogue lists that I did. Repetitiveness is really the main enemy of literary translation. In legal translation some larger segments have to be repeated, but only the ones that are identical -- not even a few words different. Writers don't usually even read other authors while working on their books, or least at the initial stages, not to get influenced by their style. Suggestions may really kill your style.

I see the uses of fuzzy matches, or suggestions in general, in technical, or medical translation. I think they are of very little use, or even a nuisance, in many other types of translation, so the use of CAT tools should never be required, because they are just something to help the translators in their work-- and it is for them only to decide if the tools help them or not to successfully complete the project.





[Edited at 2013-08-14 11:44 GMT]


 
Pages in topic:   [1 2 3] >


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


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

What are "fuzzy matches" for?

Advanced search







WordFinder Unlimited
For clarity and excellence

WordFinder is the leading dictionary service that gives you the words you want anywhere, anytime. Access 260+ dictionaries from the world's leading dictionary publishers in virtually any device. Find the right word anywhere, anytime - online or offline.

More info »
memoQ translator pro
Kilgray's memoQ is the world's fastest developing integrated localization & translation environment rendering you more productive and efficient.

With our advanced file filters, unlimited language and advanced file support, memoQ translator pro has been designed for translators and reviewers who work on their own, with other translators or in team-based translation projects.

More info »



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