Poll: What is your biggest frustration with modern Translation Management Systems (TMS) and CAT tools? Konuyu gönderen: ProZ.com Staff
|
|---|
This forum topic is for the discussion of the poll question "What is your biggest frustration with modern Translation Management Systems (TMS) and CAT tools?".
View the poll results »
| | | | Lieven Malaise 
Belçika
Local time: 00:01
Üye (2020)
Fransızca > Hollandaca
+ ...
My biggest frustration is the shift to subscription models of software in general. It apparently makes a good reason to shamelessly double, triple or even quadruple prices compared to eternal licenses.
| | | | Sebastian Witte
Almanya
Local time: 00:01
Üye (2004)
İngilizce > Almanca
+ ...
1. Paying fees for minor incremental updates. (Or rather, what Lieven said).
2. Laggy platforms. (We do not use browser-based editors when working in our own CAT tool).
3. AI features forced into them. (However, these can easily be ignored.)
Even so, we're 100% faster than when working with no CAT tool.
| | | | | Complexity and inconsisitency - an example of XTM | Jul 29 |
There are at least three to five ways to connect to XTM to manage tasks, translations, and invoicing; such complexity, cumbersomeness, and lack of consistency do not reflect high-performance engineering, but rather sheer incompetence.
And its TM and concordance research barely work well.
| | |
|
|
|
Luis M. Sosa
Ekvador
Local time: 17:01
İngilizce > İspanyolca
+ ...
Question is for both PMs and translators. But I guess only translators will show up here.
I have the impression that I already have said this in a previous related poll.
With respect to CAT tools, large projects (>15k with large TMs) turn them to virtual elephants. QA checks also become quite slow or nightmarish, and, believe me, it does not depend on the internet connection speed. Could give 3 names, but it might sound incorrect.
| | | | Zea_Mays
İtalya
Local time: 00:01
İngilizce > Almanca
+ ...
Luis M. Sosa wrote:
With respect to CAT tools, large projects (>15k with large TMs) turn them to virtual elephants. QA checks also become quite slow or nightmarish, and, believe me, it does not depend on the internet connection speed. Could give 3 names, but it might sound incorrect.
Ah, this touches a sensitive point! I often have to dismiss hundreds of false postives, and if you don't know how to do this in bulk, it can really drive you mad and cost a lot of time.
In a recent project with various 10ks words I asked the client to disable the live QA as it was slowing the editor page to prehistoric speed.
A very weak point in current CAT tools is surely the use of AI for translation. So far, none of them could beat trained neural engines, meaning you'll have tons of inconsistencies, ignored glossaries and TMs etc. etc. (AI bots do not remember things for longer periods.) I just hope that AI subscriptions will become so costly in the near term that these gimmics will disappear very soon!
[Bearbeitet am 2026-07-30 07:06 GMT]
| | | | | Stupid AI QA checks | Jul 30 |
While I am quite OK with AI translation, which makes postediting rather like proofreading with somehow more mindful checks, I am really annoyed with so many stupid AI QA checks, which unnecessarily waste much time. In the end, I don't think AI saves us a lot of time.
[Edited at 2026-07-30 07:24 GMT]
| | | | Dan Lucas
Birleşik Krallık
Local time: 23:01
Üye (2014)
Japonca > İngilizce
Zea_Mays wrote:
Ah, this touches a sensitive point! I often have to dismiss hundreds of false postives, and if you don't know how to do this in bulk, it can really drive you mad and cost a lot of time.
Now, imagine that same problem but exacerbated by the fact that you're dealing with a CJK source language, where things like unit numbers (units of ten thousand, hundred million, etc.) are different to those used in English. Neither Trados nor Phrase have sensible support for this kind of thing - I had to create my own tools to deal with these issues, and many others.
Dan
| | |
|
|
|
Zea_Mays
İtalya
Local time: 00:01
İngilizce > Almanca
+ ...
Dan Lucas wrote:
Zea_Mays wrote:
Ah, this touches a sensitive point! I often have to dismiss hundreds of false postives, and if you don't know how to do this in bulk, it can really drive you mad and cost a lot of time.
Now, imagine that same problem but exacerbated by the fact that you're dealing with a CJK source language, where things like unit numbers (units of ten thousand, hundred million, etc.) are different to those used in English. Neither Trados nor Phrase have sensible support for this kind of thing - I had to create my own tools to deal with these issues, and many others.
Dan
You'd think that's basics any CAT tool should be able to handle. Curious about your tools... In Phrase, I ended up sorting all QA flags by category or comment, marking all identical ones like in excel (mark the first entry, hold the shift key and mark the last entry) and then selecting Ignore on one of them. Gone they are! Of course, this QA madness risks to let you overlook real issues, but that's the price (I do some random checks among the flags, though).
(Glad to see you in the forums again, I think you have been greatly missed by quite some people here! )
| | | | Dan Lucas
Birleşik Krallık
Local time: 23:01
Üye (2014)
Japonca > İngilizce
Zea_Mays wrote:
You'd think that's basics any CAT tool should be able to handle.
You would, wouldn't you, and yet here we are.
Curious about your tools...
Basically xliff is text, so opening a file, extracting the content of segments, and looking at the results carefully is possible, although there are many pain points and irritating edge cases.
So, if you have something like 2億3千万円 in the source, you create a list of all the combinations that number could be when translated into English (e.g. 230 million yen, or 0.23 billion yen), then you check the target sentence for the presence of any of the numbers on that list. If the checking system doesn't see any of the numbers, it alerts the user, who then checks in the CAT tool.
Similar approaches can be used for all kinds of things, but there are many wrinkles and pitfalls for the unwary, as I know to my cost.
Dan
| | | | Chris Spurgin
Birleşik Krallık
Local time: 23:01
Rusça > İngilizce
+ ...
| Late to the party as always but.... | Jul 31 |
The CAT tools I am most familiar with still cost the same amount (and are not exactly cheap) but they don't really get any better (and don't justify the outlay) and still have many of the same bugs as they did in previous versions.
As for platform-based tools, there is one in particular which has clearly been aggressively marketed/oriented towards project managers but which is a nightmare for anyone trying to translate a document using it. PMs seem to like it but the translator func... See more The CAT tools I am most familiar with still cost the same amount (and are not exactly cheap) but they don't really get any better (and don't justify the outlay) and still have many of the same bugs as they did in previous versions.
As for platform-based tools, there is one in particular which has clearly been aggressively marketed/oriented towards project managers but which is a nightmare for anyone trying to translate a document using it. PMs seem to like it but the translator functionalities are basically nil and pretty much exist in name only.
I hope everyone is getting on well. I don't really use Proz much any more. Still keeping busy with work, though.
[Edited at 2026-07-31 13:11 GMT] ▲ Collapse
| | | | | AI Isn't Stupid. Most AI Translation Workflows Are. | 18:26 |
I often see people say that AI is not good enough for professional translation or QA.
I understand why they think that—if you're only using GPT or Claude directly, they're right. On their own, they cannot replace a professional translation workflow.
I've been using Trados and memoQ for years, but I don't rely on them for the actual translation anymore. Instead, I developed my own AI-powered translation and QA tool after more than five years of testing and refinement.
H... See more I often see people say that AI is not good enough for professional translation or QA.
I understand why they think that—if you're only using GPT or Claude directly, they're right. On their own, they cannot replace a professional translation workflow.
I've been using Trados and memoQ for years, but I don't rely on them for the actual translation anymore. Instead, I developed my own AI-powered translation and QA tool after more than five years of testing and refinement.
Here are some of the key challenges I had to solve:
1) We need a perfect SDLXliff/Xliff phrasing module
As you know we should keep tags after the translation and change status.
2) While translating humans have their translation rule. We should keep this rule while translating using gpt and claude.
We should add this on the prompt.
Our tool has all prompts for translation, editing, IIT, KMT, TGM and so on.
3) We should add glossary and tm while translating.
4) I developed fuzzy matching for the glossary and tm for 3 months.
I did not use any python library for fuzzy matching.
I developed my own module for the tm/glossary fuzzy matching.
5) My tool supports important QA such as IIT, KMT, TGM as Xbench and fix translation results from Xbench report too.
I tested this tool for more than 5 years and fixed all the bugs.
Thx. ▲ Collapse
| | | | To report site rules violations or get help, contact a site moderator: You can also contact site staff by submitting a support request » Poll: What is your biggest frustration with modern Translation Management Systems (TMS) and CAT tools? | LinguaCore | AI Translation at Your Fingertips
The underlying LLM technology of LinguaCore offers AI translations of unprecedented quality. Quick and simple. Add a human linguistic review at the end for expert-level quality at a fraction of the cost and time.
More info » |
| | CafeTran Espresso | You've never met a CAT tool this clever!
Translate faster & easier, using a sophisticated CAT tool built by a translator / developer.
Accept jobs from clients who use Trados, MemoQ, Wordfast & major CAT tools.
Download and start using CafeTran Espresso -- for free
Buy now! » |
|
| | | | X Sign in to your ProZ.com account... | | | | | |
Login to reply/comment 
