Fuzzy matching (computer-assisted translation)

Fuzzy matching, in the context of computer-assisted translation (CAT), refers to a technique used to identify segments of text in a new source document that are similar, but not identical, to segments already stored in a translation memory (TM). Unlike exact matches, which require a perfect correspondence between the source segment and a segment in the TM, fuzzy matches allow for variations, such as additions, deletions, substitutions, and transpositions of words or phrases.

The degree of similarity between the new segment and the existing TM segment is typically expressed as a percentage, often referred to as a fuzzy match score. For example, a 95% fuzzy match indicates a high degree of similarity, while a 75% fuzzy match indicates a lower degree of similarity. CAT tools use algorithms to calculate these scores, taking into account factors like edit distance and the length of the segments.

When a fuzzy match is found, the CAT tool presents the translator with the matching TM segment and highlights the differences between the source segment and the TM segment. This allows the translator to quickly adapt the existing translation, saving time and effort compared to translating the segment from scratch. The translator can then edit the suggested translation to ensure accuracy and consistency with the target language.

Fuzzy matching plays a crucial role in improving translator productivity and consistency across projects. It allows translators to leverage previously translated material, reducing redundancy and ensuring that terminology and style are consistently applied. The effectiveness of fuzzy matching depends on the quality and size of the translation memory. A larger and more well-maintained translation memory is more likely to contain relevant fuzzy matches.