How does CM Matches work?

Question

Hi all, 
 I'm hoping to get some clarity on how TM matches work and how Trados decides whether a segment is a Context Match or not. 
 Yesterday, I had a case where I was working on a project using 2 Translation Memories. When launching the pre-translation of my project, and going through the results in the Editor, I noticed something strange with segment 3. 
 Apparently, segment 3 was marked as a Context Match, however, we can see that there were 2 results found in my Translation Memories: 
 TM 1 = Employeur attractif (99%) 
 TM 2 = Un employeur attractif (99%) 
 
 But Trados used the result from TM 2 and marked the segment as Context Match. 
 
 I'm confused, as I was under the impression that as soon as several results are found in Translation Memories, the result can never be a Context Match. 
 Or am I wrong? 
 
 Would appreciate it if someone has an explanation for this. Thanks!

Paul · Answer

Matthieu Waldburger 
 This article may be worth a read... it is old and I have a feeling parts ofit may not be relevant today but it will give you a better understanding of the concepts I hope: 
 https://multifarious.filkin.com/2013/02/13/100percent/ 
 In your example the main takeaways are probably this: 
 
 you have a penalty preventing a 100% match being 100% in the TM Results window 
 since both are really 100% matches they can be considered as a potential Context Match 
 The result in TM2 must also contain additional information in the TU that makes this the preferred one in the context of your file (see the article) 
 
 Perhaps that will shed some light on things and you can be more specific if you still have questions.

Jerzy Czopik · Answer

Matthieu Waldburger 
 I do not speak and understand French, but I'll try to explain using English and German, while I suppose the situation in French will be the same. 
 Imagine you have to translate the following text: 
 You see a window. Click it to select it. You see a button. Click it to select it. 
 As you can clearly see "Click it to select it." is a repetition. So in first attempt this will be a 100% match from the TM: 
 Du siehst ein Fenster. Klicke es, um es auszuw&auml;hlen. Du siehst eine Taste. Klicke es, um es auszuw&auml;hlen. 
 Now the "Klicke es, um es auszuw&auml;hlen." does only fit for the window (Fenster, das = Neutrum). "Taste" is Femininum (Taste, die). So the second occurrence of "Click it to select it." must be translated in German to "Klicke sie, um sie auszuw&auml;hlen." 
 Now we have this order in translation: 
 Du siehst ein Fenster. Klicke es, um es auszuw&auml;hlen. Du siehst eine Taste. Klicke sie , um sie auszuw&auml;hlen. 
 And our TM contains now TWO different translations for "Click it to select it." If your next document will contain this: 
 You see a picture. Click it to select it. 
 your TM will give you a 99% match for "Click it to select it." due to multiple translations penalty. 
 But if you get this again:

You see a window. 
 Du siehst ein Fenster.

Click it to select it. 
 Klicke es, um es auszuw&auml;hlen.

You see a button. 
 Du siehst eine Taste.

Click it to select it. 
 Klicke sie, um sie auszuw&auml;hlen.

all will be context match. 
 The explanation is: The first sentence is the same as in the previous document. It has been translated exactly the same way as in the previous document. And the document structure level of both source and translation is the same, i.e. heading or list item. The next sentence is the same in source as it previously was. The previous sentence is the same as it previously was, has the same translation and the same document structure level. So TM can decide, which one of the multiple translation proposals does fit here.

Trados Studio > 1. Trados Studio

How does CM Matches work?

Top Replies