Ridiculous partial matches
Thread poster: Mikhail Kropotov

Mikhail Kropotov  Identity Verified
Russian Federation
Local time: 04:30
Member (2005)
English to Russian
+ ...
Apr 20, 2011

'An employee of a competitor'
'An Overview of the Module'

Looking at the above segments, please answer the following questions:

1. How similar are these segments?
2. If you had the translation for one of them at hand, how much would that lower the effort required to translate the other?
3. This is a 62% match in Trados. How do you like that?

Thanks for your feedback.

P.S. Yes, I noticed the post's title kills most of the sarcasm but that's forum rules for you.

[Edited at 2011-04-20 12:46 GMT]


Direct link Reply with quote
 

Manuela Ribecai  Identity Verified
Italy
Local time: 03:30
Member (2010)
English to French
+ ...
... Apr 20, 2011

"an", "of" : words with deep meaning!

XD


Direct link Reply with quote
 

Evija Rimšāne  Identity Verified
Latvia
Local time: 04:30
English to Latvian
Indeed ridiculous Apr 20, 2011

and this is why matches up to 75% are normally regarded as "no matches"

Direct link Reply with quote
 

Paul Harrison MITI
United Kingdom
Local time: 02:30
French to English
I have a whole list of these! Apr 20, 2011

I'll try and put them up later. Most are 70% matches but they do go up to 80%

Direct link Reply with quote
 

Erik Freitag  Identity Verified
Germany
Local time: 03:30
Member (2006)
Dutch to German
+ ...
agree Apr 20, 2011

Dear Mikhail,

I sympathize. Indeed, this is one of the many problems with Studio that bother me most. I've opened a thread about a similar problem recently:

http://www.proz.com/forum/cat_tools_technical_help/196156-match_algorithm_expectations.html

Kind regards,
Erik


Direct link Reply with quote
 

Selcuk Akyuz  Identity Verified
Turkey
Local time: 04:30
Member (2006)
English to Turkish
+ ...
Studio matches Apr 20, 2011

Even 85-94 matches can be useless if there are tags and numbers. So discounts only for 95-99 matches.

Direct link Reply with quote
 

ahmadwadan.com  Identity Verified
Kuwait
Local time: 04:30
English to Arabic
+ ...
Stop words Apr 21, 2011

SDL Trados should consider excluding (stop words) upon analysis. The same as search engines do.

Direct link Reply with quote
 

Grzegorz Gryc  Identity Verified
Local time: 03:30
French to Polish
+ ...
General assumptions based upon biased algorithms vs discount swindle Apr 21, 2011

Evija Rimšāne wrote:

and this is why matches up to 75% are normally regarded as "no matches"


I disagree.
I use sucessfully 45-50% match level but I use mainly DVX.
As you can see in the table kindly provided by Paul Filkin, Trados almost always gives higher match values than the competition.
http://glg.proz.com/forum/sdl_trados_support/183991-how_to_have_trados_concentrate_on_relevant_text_rather_than_tag_material_for_finding_matches.html
The fact the Trados algorithms are faulty/biased should not induce general remarks kinda "below 75% the matches make few sense" or similar.
In the most primitive scenario, a 70% match means 7 identical words in 10 word sentence which is still quite interesting for a translator.

If you speak about discounts based on the matches, it's indeed getting a big swindle now.
Even a bigger match level may be useless (e.g. see the Selcuk's opinion) while some translation offices start their discount grids from a Trados or Trados-like 50% level...

GG

[Edited at 2011-04-21 12:31 GMT]


Direct link Reply with quote
 

Grzegorz Gryc  Identity Verified
Local time: 03:30
French to Polish
+ ...
Some stats... Apr 21, 2011

Mikhail Kropotov wrote:

'An employee of a competitor'
'An Overview of the Module'

Looking at the above segments, please answer the following questions:

1. How similar are these segments?

According to the basic Levenshtein distance algorithm, something like 25-28%.

2. If you had the translation for one of them at hand, how much would that lower the effort required to translate the other?

Well, maybe 10-15%.

3. This is a 62% match in Trados. How do you like that?

It's a well known nonsense.
The problem with Trados is it doesn't consider the word length in its algorithms.
As the grammatical words (articles, prepositions etc.) are normally shorter, they're normally overweighted and the translarors are underpaid.

This flaw is not only Trados related but the Trados market share legitimates it, i.e. a lot of people think a generalized nonsense is truth.

Cheers
GG


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 »

Ridiculous partial matches

Advanced search







SDL Trados Studio 2017 Freelance
The leading translation software used by over 250,000 translators.

SDL Trados Studio 2017 helps translators increase translation productivity whilst ensuring quality. Combining translation memory, terminology management and machine translation in one simple and easy-to-use environment.

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 »



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