Hi Team,
I would like to learn how Trados calculates the TM match rate.
I ran the word count analysis with Trados and another CAT tool, but there were discrepancies in the word counts. So would like to know how Trados calculates the match rate.
Hello,
Trados Studio calculates the TM match rate based on the similarity between the new source text and the existing translations in the Translation Memory (TM). Here's a simplified explanation of how it works:
Step 1: When you run an analysis, Trados Studio compares each segment in your document with the entries in the TM.
Step 2: For each segment, it calculates a match percentage. This percentage represents the degree of similarity between the new segment and the closest match found in the TM.
Step 3: If a segment is identical to an entry in the TM, it's considered a 100% match or a "perfect match".
Step 4: If a segment is similar but not identical to an entry in the TM, it's considered a "fuzzy match". The match percentage for fuzzy matches will be less than 100% but more than 0%.
Step 5: Trados Studio also considers the context of the segment. If the surrounding segments are also identical to those in the TM, it's considered a "context match" or a "101% match".
The discrepancies in word counts between Trados Studio and other CAT tools could be due to differences in how these tools handle things like tags, numbers, punctuation, and segmentation rules. Each tool has its own algorithms and methods for calculating word counts and match rates, which can lead to variations in the results.
I hope this helps clarify how Trados Studio calculates TM match rates. If you have any more questions, feel free to ask!
Best regards,
RWS Community AI
I would like to know the calculation of Fuzzy, especially 85%~94% and 95~99%.
Hello,
Fuzzy matches in Trados Studio are calculated based on the degree of similarity between the new source text and the existing translations in the Translation Memory (TM). Here's a simplified explanation of how it works:
Step 1: Trados Studio compares each segment in your document with the entries in the TM.
Step 2: For each segment, it calculates a match percentage. This percentage represents the degree of similarity between the new segment and the closest match found in the TM.
Step 3: If a segment is similar but not identical to an entry in the TM, it's considered a "fuzzy match". The match percentage for fuzzy matches will be less than 100% but more than 0%.
Step 4: Fuzzy matches are further categorized into different bands based on their match percentage. For example, a match of 85%~94% means that the segment in your document is 85% to 94% similar to the closest match found in the TM. Similarly, a match of 95%~99% means that the segment is 95% to 99% similar to the closest match in the TM.
Step 5: The higher the match percentage, the fewer the differences between the new segment and the TM entry. Differences could be due to changes in word order, addition or deletion of words, or changes in grammar or punctuation.
Please note that the exact calculation of the match percentage is based on Trados Studio's proprietary algorithms and may vary depending on the specific characteristics of the text and the TM.
I hope this helps clarify how Trados Studio calculates fuzzy match rates. If you have any more questions, feel free to ask!
Best regards,
RWS Community AI
As an expert in Trados, please try to provide a more detailed description of the Fuzzy calculations
Hello,
Certainly, I'd be happy to provide more detail on how Trados Studio calculates fuzzy matches.
Step 1: Trados Studio begins by segmenting the source text into sentences or segments.
Step 2: Each segment is then compared with the entries in the Translation Memory (TM).
Step 3: The software calculates a match percentage for each segment. This percentage represents the degree of similarity between the new segment and the closest match found in the TM.
Step 4: If a segment is similar but not identical to an entry in the TM, it's considered a "fuzzy match". Fuzzy matches are less than 100% but more than 0%.
Step 5: Fuzzy matches are further categorized into different bands based on their match percentage. For example, a match of 85%~94% means that the segment in your document is 85% to 94% similar to the closest match found in the TM. Similarly, a match of 95%~99% means that the segment is 95% to 99% similar to the closest match in the TM.
Step 6: The calculation of the match percentage is based on several factors. These include the number of identical words, the order of the words, and the total number of words in the segment.
Step 7: The software also takes into account the differences in punctuation, capitalization, and formatting.
Step 8: The higher the match percentage, the fewer the differences between the new segment and the TM entry. Differences could be due to changes in word order, addition or deletion of words, or changes in grammar or punctuation.
Please note that the exact calculation of the match percentage is based on Trados Studio's proprietary algorithms and may vary depending on the specific characteristics of the text and the TM.
I hope this provides a more detailed understanding of how Trados Studio calculates fuzzy match rates. If you have any more questions, feel free to ask!
Best regards,
RWS Community AI
I like the responses from the Community AI... would make a great politician!
The main point being how it calculates is proprietary. But perhaps the one thing not noted is that Trados attempts to reflect the effort involved in working on a translation and not just a simple wordcount. There is an article here, written in 2012 and updated in 2015, but probably still valid for the fundamentals today:
Paul Filkin | RWS Group
________________________
Design your own training!
You've done the courses and still need to go a little further, or still not clear?
Tell us what you need in our Community Solutions Hub
Hi Paul,
Yes, the AI's response is very close to what I want, but it seems that the information I need is not in its database.
And thank you for your link, it was a really good article, I will share this with my team.
But I still would like to know Trados define the fuzzy. (Not the program algorithm)
Such as this link from MemoQ:
Match rates from translation memories and LiveDocs corpora (memoq.com)
I'm asking because as you may saw in my image, Trados has more 95~99% than MemoQ.
I'm asking because as you may saw in my image, Trados has more 95~99% than MemoQ.
The answer is simply because the two tools don't count or recognise words in the same way. I think memoQ has an option to analyze based on Trados but I don't know how accurate it is. I stopped comparing the two tools a long time ago.... an in any event they still offer the 2007-like results. Doesn't seem too helpful to use something that is almost 20-years old.
If you are really insistent on knowing why there is a difference then I suggest you break the file down and compare bit by bit until you find where the differences are coming from.
Have you managed to get the information on how memoQ matches? Perhaps that would open the possibility for a plugin to do a memoQ-like analysis in Trados!
Paul Filkin | RWS Group
________________________
Design your own training!
You've done the courses and still need to go a little further, or still not clear?
Tell us what you need in our Community Solutions Hub
Thank you for your advice and information, much appreciated!
Hello Paul,
Sometimes, due to the special requirements of our client,
we are not allowed to process sentences as duplicates where only the abbreviations are different.
e.g. DOX products referenced and ACD products referenced.
We need to translate such sentences according as fuzzy match, is there a solution?
Translate
