Fuzzy Match
A fuzzy match is a translation memory (TM) suggestion where the stored source segment is similar — but not identical — to the current source text being translated. The similarity is expressed as a percentage: 100% means an exact match, while 75–99% indicates a fuzzy match.
CAT tools highlight fuzzy matches so translators can adapt them rather than translating from scratch. A 95% fuzzy match might differ only by a number, product name, or punctuation change — easy to edit. A 70% fuzzy match might require substantial rewriting. Agencies typically set pricing tiers based on fuzzy match bands: exact matches (100%) are priced lowest because they require no editing; 75–84% matches are priced higher; below 70% are often priced as full new translation.
Fuzzy match thresholds are configurable. Most systems set a minimum threshold of 50–70% below which matches are not shown, to avoid showing suggestions that are more confusing than helpful.
Key concepts
- Percentage score = similarity between stored TM segment and current source text
- 100% = exact match; below ~70% = generally not useful enough to show
- Agencies price translation by fuzzy match band — higher match % = lower per-word rate
How Vernacia uses this
Vernacia uses translation memory to find exact and fuzzy matches before sending segments to the AI model. High-confidence matches are reused directly; lower-confidence matches inform the AI prompt context.
Related terms
Translate PPTX, PDF, DOCX, and 11 more formats — free to start.
Try Vernacia free