Strange (mis-)calculation of matches in XTM
Thread poster: ORSZAGH Technologies

ORSZAGH Technologies
Cyprus
Local time: 08:39
Czech to Slovak
+ ...
Apr 14, 2015

Hello, can anyone from XTM please explain my below findings?

String in TM: Follow these guidelines when selecting system components:
String in file for translation: Follow these guidelines when selecting system components for either initial implant or replacement:

XTM match value: 72%
Trados 2007 match value: 55%

or this one:

String in TM: electrocardiograms (ECG)
String in file for translation: Effect on electrocardiograms (ECGs)
XTM match value: 74%
Trados 2007 match value: 54%

So if I have a string with 4 words and get 2 of them from the TM, how can this be a 74% match with me still having to supply the other 2?

Thank you.


 

XTM Intl
United Kingdom
Local time: 06:39
RE: Apr 14, 2015

Hello,

Matching algorithm goes though all characters in the phrases and count match percentage based on overall number of compatible characters not only the number of words.

Kind regards,
Przemysław Nowak (XTM Support)


 

Jaroslaw Michalak  Identity Verified
Poland
Local time: 07:39
Member (2004)
English to Polish
Rate difference? Apr 14, 2015

In other words, if you assume your regular rates for projects done in XTM, you will earn much less for fuzzy matches than in case of other tools?

[Edited at 2015-04-14 19:55 GMT]


 

ORSZAGH Technologies
Cyprus
Local time: 08:39
Czech to Slovak
+ ...
TOPIC STARTER
earn less with XTM Apr 15, 2015

Jabberwock wrote:

In other words, if you assume your regular rates for projects done in XTM, you will earn much less for fuzzy matches than in case of other tools?

[Edited at 2015-04-14 19:55 GMT]


Yes, this is what I have found out. The problem is not only that you will earn less but also that this calculation method has already gone way too far in the effort to lower the effective word count and now it is plain unfair and does not correspond to the reality.


 

ORSZAGH Technologies
Cyprus
Local time: 08:39
Czech to Slovak
+ ...
TOPIC STARTER
Character based word counts? Apr 15, 2015

XTM Intl wrote:

Hello,

Matching algorithm goes though all characters in the phrases and count match percentage based on overall number of compatible characters not only the number of words.

Kind regards,
Przemysław Nowak (XTM Support)


Hello Przemysław,

Many thanks for getting back to me regarding this serious issue. I find this approach quite incorrect and unfair since it basically means that XTM automatically assumes that if a word is shorter in English, it will be shorter in the target language, too, which is of course not true. I had to increase rates for projects analyzed in XTM because of this issue and as soon as others start to realize this, it can turn against XTM in the end. XTM is applying language analysis tools that should not be applied to translation tasks and which are only applicable to analyzing single-language texts when no translation is involved. Unlike standard methods, this approach can not be used to measure the effort involved in translation.

Just a side note, I was told that XTM also uses the Levenshtein distance (http://en.wikipedia.org/wiki/Levenshtein_distance) to calculate matches, is this really true? If so, this is also unacceptable since while the Levenshtein distance between e.g. "kitten" and "sitting" is 3 (since three edits can change one into the other, and there is no way to do it with fewer than three edits), these words are translated completely differently in the target language and so this metrics is irrelevant and does not say anything about the effort needed to produce the translation. Please advise.

Thank you,
Lubo



[Edited at 2015-04-15 09:31 GMT]


 


There is no moderator assigned specifically to this forum.
To report site rules violations or get help, please contact site staff »


Strange (mis-)calculation of matches in XTM

Advanced search






SDL Trados Studio 2019 Freelance
The leading translation software used by over 250,000 translators.

SDL Trados Studio 2019 has evolved to bring translators a brand new experience. Designed with user experience at its core, Studio 2019 transforms how new users get up and running, helps experienced users make the most of the powerful features, ensures new

More info »
SDL MultiTerm 2019
Guarantee a unified, consistent and high-quality translation with terminology software by the industry leaders.

SDL MultiTerm 2019 allows translators to create one central location to store and manage multilingual terminology, and with SDL MultiTerm Extract 2019 you can automatically create term lists from your existing documentation to save time.

More info »



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