Fuzzy matches or FMs are segments that have been translated before (in your current or in previous segments), more or less …
This means: The segments haven't been translated exactly before (then we would have Exact Matches) but a similar segment has been translated before.
Similarity can mean:
- The same segment text, but with different punctuation marks
- The same segment text, but with different formatting
- Almost the same segment text, with a typo
- Almost the same segment text, with an addition (word, space, letter)
- Almost the same segment text, minus a string (word, space, letter)
- etc.
You can influence CafeTran's handling of FMs via Edit > Options > Memory:
- Fuzzy match insert threshold (%)
- The field lets you define the accuracy of the best fuzzy match at which the program inserts it automatically into the target language box.
- Fuzzy match threshold (%)
- Sets the minimum accuracy of the fuzzy matches which are displayed by CafeTran.
- Fuzzy match display limit:
- Sets the maximum number of the fuzzy matches displayed by CafeTran.
You can manually insert FMs via Translation > Segment matches > Insert Fuzzy match 1 / 2 /3.
You can let CafeTran show a list of all available FMs by pressing F1.
Using Fuzzy Auto-correction
Fuzzy matches can be corrected automatically by applying exact subsegment matches to their missing (incorrect) parts. For example, imagine that you have the following English-Spanish translation unit in your translation memory or termbase:
I like the red car = Me gusta el coche rojo
Later, when you translate the similar English sentence such as:
I like the blue car (Me gusta el coche azul)
CafeTran will retrieve the following fuzzy match from your translation memory or termbase:
‘Me gusta el coche rojo’
With the Fuzzy AutoCorrector function on, the program can substitute the missing word ‘azul’ for the word ‘rojo’, provided that:
- it finds the translation for ‘blue’ (blue = azul) in your translation memories or termbases, and/or
- it finds the translation for ‘red’ (red = rojo) in in your translation memories or termbases.
The Fuzzy AutoCorrector can be turned on/off by checking/unchecking ‘Fuzzy Auto-correction’ at Translation > Options.
Fuzzy matches and tags
CafeTran doesn't insert tags in FMs. Perhaps this will be added in a later build.
Automatic repair of fuzzy matches
When you have settings like these:
CafeTran can use a previously translated fragment to repair a fuzzy match. Here the translation for ‘simple’ is derived from the TM (project) and inserted at the correct place, which is a ‘fuzzy match repair’:
