TM matches
When Trados Studio performs a lookup, it compares the content of the source file segment against segments of the same language in the translation memory (TM). It looks for TM segments which have enough content in common with the file segment for the segment translation held in the TM.
The degree of match between the source file segment and a TM segment is expressed as a percentage figure. For example, if the content of a TM segment matches the file segment exactly, that is a 100% match. To qualify as a 100% match, the entire content (all characters and character formatting) of both the source file segment and the TM segment must match exactly.
A context match is an exact match between a document segment and a TM segment match, occurring in the same file context.
A fuzzy match is a less than 100% match. In theory, a fuzzy match could be anything between 99% and 1%. However, there is a point below which the content match is insufficient, which makes the content unusable. This cut-off point is known as the 'fuzzy match threshold'.
Fuzzy match threshold
The default fuzzy match threshold for Trados Studio is 70%. This means that when the degree of match between a source file segment and a TM segment is less than 70%, they are treated as having no match at all. The fuzzy match threshold is a setting that can be raised or lowered.
To set the fuzzy match threshold for your:
- Project - Use the Project Settings dialog.
- Active document - Use the Project Settings dialog.
- Default settings - Use the Options dialog.
Context matches
A context match is better than a 100% match. To be a context match, the TM segment must be a 100% match to the file segment and the two segments must have the same file context. For the file segment and the TM segment to have the same context, they must both have been preceded by the same segment.
Context matching is possible because context information is stored in the TM. When you add a new translation to a TM, three segments are added:
- The source file segment
- The segment translation
- The segment preceding the source file segment. If there is no preceding segment, other context information is uploaded, such as information that the segment was the file header, for example.
Context segments cannot be seen when you view the contents of the TM.
The context information is reliant on placeables, so any changes made to the recognition settings of your TM will invalidate the context match. The only way you could get context match back would be to return to your original recognition settings or retranslate the segments.
PerfectMatch
A PerfectMatch is a form of context match that compares updated source files to a corresponding set of existing bilingual files rather than to a TM. Segment matches, known as PerfectMatches, are checked for context, that is, the surrounding entries are checked to ensure that they are the same. In addition, if your existing bilingual file was segmented differently to the current file or had merged segments, PerfectMatch can dynamically merge up to three consecutive segments in your current file to improve your match results. The translations are then extracted from the existing bilingual files and transferred to the updated source files.
Because the PerfectMatch segment matching process includes a check for context and segmentation, PerfectMatch units typically need no further translation or editing during translation.