MateCat killed the tags
Thread poster: MateCat Support
When working with MateCat, you find a clean and simple interface. It lets you read the text effortlessly and focus on its meaning to complete your translation fast. We achieved that by getting rid of the clutter you find in traditional CAT tools. No unnecessary buttons, icons or extras when you don’t need them.
Now we’ve taken it a step further and removed the tags.
Tags slow you down while reading and understanding the source sentence. When typing the translation, tags force you to interrupt your cognitive process to decide what to do with them. Removing tags will help you deliver better quality translations in less time.
We analysed millions of segments processed by professionals like you. We observed how they handled all types of tags and used Machine Learning techniques to let MateCat learn where to place them.
Find out how it works:
| | CafeTran Training
Local time: 12:12
| Sounds good, but ... || Oct 8, 2016 |
MateCat Support wrote:
Tags slow you down while reading and understanding the source sentence.
Yes, this can be a problem. And bless you for trying to find a solution. However, after having watched your video, I'm not sure whether your approach is an improvement.
Excuse me if I'm wrong, but aren't you just postponing the moment of tag checking until the moment where a translator wants to proceed to the next segment?
How distracting ist that?
Instead of just pressing the keyboard shortcut (clicking the icon) to proceed to the next segment, she'll have to click Check tags first.
Once I've finished the translation of a segment, I cannot wait to see the content of the next segment (eager as I am ). I don't want to be bothered with pressing some extra keys and checking any tags first.
But perhaps it's just me, sticking to old habits?
| Good point but... || Oct 8, 2016 |
Thanks for your reply and providing a good point. I believe some translators will agree with you, whether for old habits or because they prefer a different approach. The feature is on by default but can be disabled by the user.
It is true that translators still need to check the tags, but we've observed that it is much faster to let the machine place the tags where they belong and then confirm the position rather than having to place them manually (even with shortcuts or other tricks).
From our tests, it seemed that the translators were more distracted by finding the tags in the source segment and having to think where to place them while writing the translation. This interrupted the flow of words when creating the target segment.
Our goal is going toward a solution where translators will not have to confirm the tags. To do that, we'll keep improving the system that places the tags automatically until we feel confident about having the machine do the job without the translators even seeing that. At that stage, the tags will be placed in the document when downloading the translation.
We'll probably get there, but we believe that it is important to leave to the professionals the option to check and fix what the machine does. After all, the machine is not as smart as well-trained translators are.
| || || |
There is no moderator assigned specifically to this forum.
To report site rules violations or get help, please contact site staff »